Google Git
Sign in
webkit / WebKit / 8c216134dc325a54db94b409bbb47f6436cd0c1f / . / WebCore / rendering / bidi.cpp
blob: a27fa62d82783f67e1ddfd2103073024a4faa5f0 [file] [log] [blame]
/**
* This file is part of the html renderer for KDE.
*
* Copyright (C) 2000 Lars Knoll (knoll@kde.org)
* Copyright (C) 2004, 2006 Apple Computer, Inc.
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Library General Public
* License as published by the Free Software Foundation; either
* version 2 of the License, or (at your option) any later version.
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Library General Public License for more details.
*
* You should have received a copy of the GNU Library General Public License
* along with this library; see the file COPYING.LIB. If not, write to
* the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
* Boston, MA 02111-1307, USA.
*
*/
#include "config.h"
#include "bidi.h"
#include "Document.h"
#include "Element.h"
#include "FrameView.h"
#include "InlineTextBox.h"
#include "RenderArena.h"
#include "RenderView.h"
#include "break_lines.h"
#include <wtf/AlwaysInline.h>
#include <wtf/Vector.h>
using namespace std;
namespace WebCore {
// an iterator which traverses all the objects within a block
struct BidiIterator {
BidiIterator() : block(0), obj(0), pos(0) {}
BidiIterator(RenderBlock* b, RenderObject* o, unsigned int p)
: block(b), obj(o), pos(p) {}
void increment(BidiState& state);
bool atEnd() const;
UChar current() const;
WTF::Unicode::Direction direction() const;
RenderBlock* block;
RenderObject* obj;
unsigned int pos;
};
struct BidiState {
BidiState() : context(0), dir(WTF::Unicode::OtherNeutral), adjustEmbedding(false), reachedEndOfLine(false) {}
BidiIterator sor;
BidiIterator eor;
BidiIterator last;
BidiIterator current;
RefPtr<BidiContext> context;
BidiStatus status;
WTF::Unicode::Direction dir;
bool adjustEmbedding;
BidiIterator endOfLine;
bool reachedEndOfLine;
BidiIterator lastBeforeET;
};
inline bool operator==(const BidiStatus& status1, const BidiStatus& status2)
{
return status1.eor == status2.eor && status1.last == status2.last && status1.lastStrong == status2.lastStrong;
}
inline bool operator!=(const BidiStatus& status1, const BidiStatus& status2)
{
return !(status1 == status2);
}
// Used to track a list of chained bidi runs.
static BidiRun* sFirstBidiRun;
static BidiRun* sLastBidiRun;
static BidiRun* sLogicallyLastBidiRun;
static int sBidiRunCount;
static BidiRun* sCompactFirstBidiRun;
static BidiRun* sCompactLastBidiRun;
static int sCompactBidiRunCount;
static bool sBuildingCompactRuns;
// Midpoint globals. The goal is not to do any allocation when dealing with
// these midpoints, so we just keep an array around and never clear it. We track
// the number of items and position using the two other variables.
static Vector<BidiIterator>* smidpoints;
static unsigned sNumMidpoints;
static unsigned sCurrMidpoint;
static bool betweenMidpoints;
static bool isLineEmpty = true;
static bool previousLineBrokeCleanly = true;
static bool emptyRun = true;
static int numSpaces;
static void embed(WTF::Unicode::Direction, BidiState&);
static void appendRun(BidiState&);
static int getBPMWidth(int childValue, Length cssUnit)
{
if (!cssUnit.isIntrinsicOrAuto())
return (cssUnit.isFixed() ? cssUnit.value() : childValue);
return 0;
}
static int getBorderPaddingMargin(RenderObject* child, bool endOfInline)
{
RenderStyle* cstyle = child->style();
int result = 0;
bool leftSide = (cstyle->direction() == LTR) ? !endOfInline : endOfInline;
result += getBPMWidth((leftSide ? child->marginLeft() : child->marginRight()),
(leftSide ? cstyle->marginLeft() :
cstyle->marginRight()));
result += getBPMWidth((leftSide ? child->paddingLeft() : child->paddingRight()),
(leftSide ? cstyle->paddingLeft() :
cstyle->paddingRight()));
result += leftSide ? child->borderLeft() : child->borderRight();
return result;
}
static int inlineWidth(RenderObject* child, bool start = true, bool end = true)
{
int extraWidth = 0;
RenderObject* parent = child->parent();
while (parent->isInline() && !parent->isInlineBlockOrInlineTable()) {
if (start && parent->firstChild() == child)
extraWidth += getBorderPaddingMargin(parent, false);
if (end && parent->lastChild() == child)
extraWidth += getBorderPaddingMargin(parent, true);
child = parent;
parent = child->parent();
}
return extraWidth;
}
#ifndef NDEBUG
static bool inBidiRunDestroy;
#endif
void BidiRun::destroy(RenderArena* renderArena)
{
#ifndef NDEBUG
inBidiRunDestroy = true;
#endif
delete this;
#ifndef NDEBUG
inBidiRunDestroy = false;
#endif
// Recover the size left there for us by operator delete and free the memory.
renderArena->free(*(size_t *)this, this);
}
void* BidiRun::operator new(size_t sz, RenderArena* renderArena) throw()
{
return renderArena->allocate(sz);
}
void BidiRun::operator delete(void* ptr, size_t sz)
{
assert(inBidiRunDestroy);
// Stash size where destroy() can find it.
*(size_t*)ptr = sz;
}
static void deleteBidiRuns(RenderArena* arena)
{
emptyRun = true;
if (!sFirstBidiRun)
return;
BidiRun* curr = sFirstBidiRun;
while (curr) {
BidiRun* s = curr->nextRun;
curr->destroy(arena);
curr = s;
}
sFirstBidiRun = 0;
sLastBidiRun = 0;
sBidiRunCount = 0;
}
// ---------------------------------------------------------------------
/* a small helper class used internally to resolve Bidi embedding levels.
Each line of text caches the embedding level at the start of the line for faster
relayouting
*/
BidiContext::BidiContext(unsigned char l, WTF::Unicode::Direction e, BidiContext *p, bool o)
: level(l), override(o), m_dir(e)
{
parent = p;
if (p) {
p->ref();
m_basicDir = p->basicDir();
} else
m_basicDir = e;
count = 0;
}
BidiContext::~BidiContext()
{
if (parent)
parent->deref();
}
void BidiContext::ref() const
{
count++;
}
void BidiContext::deref() const
{
count--;
if (count <= 0)
delete this;
}
bool operator==(const BidiContext& c1, const BidiContext& c2)
{
if (&c1 == &c2)
return true;
if (c1.level != c2.level || c1.override != c2.override || c1.dir() != c2.dir() || c1.basicDir() != c2.basicDir())
return false;
if (!c1.parent)
return !c2.parent;
return c2.parent && *c1.parent == *c2.parent;
}
inline bool operator!=(const BidiContext& c1, const BidiContext& c2)
{
return !(c1 == c2);
}
// ---------------------------------------------------------------------
inline bool operator==(const BidiIterator& it1, const BidiIterator& it2)
{
if (it1.pos != it2.pos)
return false;
if (it1.obj != it2.obj)
return false;
return true;
}
inline bool operator!=(const BidiIterator& it1, const BidiIterator& it2)
{
if (it1.pos != it2.pos)
return true;
if (it1.obj != it2.obj)
return true;
return false;
}
static inline RenderObject* bidiNext(RenderBlock* block, RenderObject* current, BidiState& bidi,
bool skipInlines = true, bool* endOfInline = 0)
{
RenderObject* next = 0;
bool oldEndOfInline = endOfInline ? *endOfInline : false;
if (endOfInline)
*endOfInline = false;
while (current) {
if (!oldEndOfInline && !current->isFloating() && !current->isReplaced() && !current->isPositioned()) {
next = current->firstChild();
if (next && bidi.adjustEmbedding && next->isInlineFlow()) {
EUnicodeBidi ub = next->style()->unicodeBidi();
if (ub != UBNormal) {
TextDirection dir = next->style()->direction();
WTF::Unicode::Direction d = (ub == Embed
? (dir == RTL ? WTF::Unicode::RightToLeftEmbedding : WTF::Unicode::LeftToRightEmbedding)
: (dir == RTL ? WTF::Unicode::RightToLeftOverride : WTF::Unicode::LeftToRightOverride));
embed(d, bidi);
}
}
}
if (!next) {
if (!skipInlines && !oldEndOfInline && current->isInlineFlow()) {
next = current;
if (endOfInline)
*endOfInline = true;
break;
}
while (current && current != block) {
if (bidi.adjustEmbedding && current->isInlineFlow() && current->style()->unicodeBidi() != UBNormal)
embed(WTF::Unicode::PopDirectionalFormat, bidi);
next = current->nextSibling();
if (next) {
if (bidi.adjustEmbedding && next->isInlineFlow()) {
EUnicodeBidi ub = next->style()->unicodeBidi();
if (ub != UBNormal) {
TextDirection dir = next->style()->direction();
WTF::Unicode::Direction d = (ub == Embed
? (dir == RTL ? WTF::Unicode::RightToLeftEmbedding: WTF::Unicode::LeftToRightEmbedding)
: (dir == RTL ? WTF::Unicode::RightToLeftOverride : WTF::Unicode::LeftToRightOverride));
embed(d, bidi);
}
}
break;
}
current = current->parent();
if (!skipInlines && current && current != block && current->isInlineFlow()) {
next = current;
if (endOfInline)
*endOfInline = true;
break;
}
}
}
if (!next)
break;
if (next->isText() || next->isBR() || next->isFloating() || next->isReplaced() || next->isPositioned()
|| ((!skipInlines || !next->firstChild()) // Always return EMPTY inlines.
&& next->isInlineFlow()))
break;
current = next;
}
return next;
}
static RenderObject* bidiFirst(RenderBlock* block, BidiState& bidi, bool skipInlines = true )
{
if (!block->firstChild())
return 0;
RenderObject* o = block->firstChild();
if (o->isInlineFlow()) {
if (bidi.adjustEmbedding) {
EUnicodeBidi ub = o->style()->unicodeBidi();
if (ub != UBNormal) {
TextDirection dir = o->style()->direction();
WTF::Unicode::Direction d = (ub == Embed
? (dir == RTL ? WTF::Unicode::RightToLeftEmbedding : WTF::Unicode::LeftToRightEmbedding)
: (dir == RTL ? WTF::Unicode::RightToLeftOverride : WTF::Unicode::LeftToRightOverride));
embed(d, bidi);
}
}
if (skipInlines && o->firstChild())
o = bidiNext(block, o, bidi, skipInlines);
else
return o; // Never skip empty inlines.
}
if (o && !o->isText() && !o->isBR() && !o->isReplaced() && !o->isFloating() && !o->isPositioned())
o = bidiNext(block, o, bidi, skipInlines);
return o;
}
inline void BidiIterator::increment(BidiState& bidi)
{
if (!obj)
return;
if (obj->isText()) {
pos++;
if (pos >= static_cast<RenderText *>(obj)->stringLength()) {
obj = bidiNext(block, obj, bidi);
pos = 0;
}
} else {
obj = bidiNext(block, obj, bidi);
pos = 0;
}
}
inline bool BidiIterator::atEnd() const
{
return !obj;
}
UChar BidiIterator::current() const
{
if (!obj || !obj->isText())
return 0;
RenderText* text = static_cast<RenderText*>(obj);
if (!text->text())
return 0;
return text->text()[pos];
}
ALWAYS_INLINE WTF::Unicode::Direction BidiIterator::direction() const
{
if (!obj)
return WTF::Unicode::OtherNeutral;
if (obj->isListMarker())
return obj->style()->direction() == LTR ? WTF::Unicode::LeftToRight : WTF::Unicode::RightToLeft;
if (!obj->isText())
return WTF::Unicode::OtherNeutral;
RenderText* renderTxt = static_cast<RenderText*>(obj);
if (pos >= renderTxt->stringLength())
return WTF::Unicode::OtherNeutral;
return WTF::Unicode::direction(renderTxt->text()[pos]);
}
// -------------------------------------------------------------------------------------------------
static void addRun(BidiRun* bidiRun)
{
if (!sFirstBidiRun)
sFirstBidiRun = sLastBidiRun = bidiRun;
else {
sLastBidiRun->nextRun = bidiRun;
sLastBidiRun = bidiRun;
}
sBidiRunCount++;
bidiRun->compact = sBuildingCompactRuns;
// Compute the number of spaces in this run,
if (bidiRun->obj && bidiRun->obj->isText()) {
RenderText* text = static_cast<RenderText*>(bidiRun->obj);
if (text->text()) {
for (int i = bidiRun->start; i < bidiRun->stop; i++) {
UChar c = text->text()[i];
if (c == ' ' || c == '\n' || c == '\t')
numSpaces++;
}
}
}
}
static void reverseRuns(int start, int end)
{
if (start >= end)
return;
assert(start >= 0 && end < sBidiRunCount);
// Get the item before the start of the runs to reverse and put it in
// |beforeStart|. |curr| should point to the first run to reverse.
BidiRun* curr = sFirstBidiRun;
BidiRun* beforeStart = 0;
int i = 0;
while (i < start) {
i++;
beforeStart = curr;
curr = curr->nextRun;
}
BidiRun* startRun = curr;
while (i < end) {
i++;
curr = curr->nextRun;
}
BidiRun* endRun = curr;
BidiRun* afterEnd = curr->nextRun;
i = start;
curr = startRun;
BidiRun* newNext = afterEnd;
while (i <= end) {
// Do the reversal.
BidiRun* next = curr->nextRun;
curr->nextRun = newNext;
newNext = curr;
curr = next;
i++;
}
// Now hook up beforeStart and afterEnd to the newStart and newEnd.
if (beforeStart)
beforeStart->nextRun = endRun;
else
sFirstBidiRun = endRun;
startRun->nextRun = afterEnd;
if (!afterEnd)
sLastBidiRun = startRun;
}
static void chopMidpointsAt(RenderObject* obj, unsigned pos)
{
if (!sNumMidpoints)
return;
BidiIterator* midpoints = smidpoints->data();
for (unsigned i = 0; i < sNumMidpoints; i++) {
const BidiIterator& point = midpoints[i];
if (point.obj == obj && point.pos == pos) {
sNumMidpoints = i;
break;
}
}
}
static void checkMidpoints(BidiIterator& lBreak, BidiState& bidi)
{
// Check to see if our last midpoint is a start point beyond the line break. If so,
// shave it off the list, and shave off a trailing space if the previous end point doesn't
// preserve whitespace.
if (lBreak.obj && sNumMidpoints && sNumMidpoints%2 == 0) {
BidiIterator* midpoints = smidpoints->data();
BidiIterator& endpoint = midpoints[sNumMidpoints-2];
const BidiIterator& startpoint = midpoints[sNumMidpoints-1];
BidiIterator currpoint = endpoint;
while (!currpoint.atEnd() && currpoint != startpoint && currpoint != lBreak)
currpoint.increment(bidi);
if (currpoint == lBreak) {
// We hit the line break before the start point. Shave off the start point.
sNumMidpoints--;
if (endpoint.obj->style()->collapseWhiteSpace()) {
if (endpoint.obj->isText()) {
// Don't shave a character off the endpoint if it was from a soft hyphen.
RenderText* textObj = static_cast<RenderText*>(endpoint.obj);
if (endpoint.pos+1 < textObj->length()) {
if (textObj->text()[endpoint.pos+1] == SOFT_HYPHEN)
return;
} else if (startpoint.obj->isText()) {
RenderText *startText = static_cast<RenderText*>(startpoint.obj);
if (startText->length() > 0 && startText->text()[0] == SOFT_HYPHEN)
return;
}
}
endpoint.pos--;
}
}
}
}
static void addMidpoint(const BidiIterator& midpoint)
{
if (!smidpoints)
return;
if (smidpoints->size() <= sNumMidpoints)
smidpoints->resize(sNumMidpoints + 10);
BidiIterator* midpoints = smidpoints->data();
midpoints[sNumMidpoints++] = midpoint;
}
static void appendRunsForObject(int start, int end, RenderObject* obj, BidiState &bidi)
{
if (start > end || obj->isFloating() ||
(obj->isPositioned() && !obj->hasStaticX() && !obj->hasStaticY() && !obj->container()->isInlineFlow()))
return;
bool haveNextMidpoint = (smidpoints && sCurrMidpoint < sNumMidpoints);
BidiIterator nextMidpoint;
if (haveNextMidpoint)
nextMidpoint = smidpoints->at(sCurrMidpoint);
if (betweenMidpoints) {
if (!(haveNextMidpoint && nextMidpoint.obj == obj))
return;
// This is a new start point. Stop ignoring objects and
// adjust our start.
betweenMidpoints = false;
start = nextMidpoint.pos;
sCurrMidpoint++;
if (start < end)
return appendRunsForObject(start, end, obj, bidi);
}
else {
if (!smidpoints || !haveNextMidpoint || (obj != nextMidpoint.obj)) {
addRun(new (obj->renderArena()) BidiRun(start, end, obj, bidi.context.get(), bidi.dir));
return;
}
// An end midpoint has been encountered within our object. We
// need to go ahead and append a run with our endpoint.
if (int(nextMidpoint.pos+1) <= end) {
betweenMidpoints = true;
sCurrMidpoint++;
if (nextMidpoint.pos != UINT_MAX) { // UINT_MAX means stop at the object and don't include any of it.
if (int(nextMidpoint.pos+1) > start)
addRun(new (obj->renderArena())
BidiRun(start, nextMidpoint.pos+1, obj, bidi.context.get(), bidi.dir));
return appendRunsForObject(nextMidpoint.pos+1, end, obj, bidi);
}
}
else
addRun(new (obj->renderArena()) BidiRun(start, end, obj, bidi.context.get(), bidi.dir));
}
}
static void appendRun(BidiState &bidi)
{
if (emptyRun || !bidi.eor.obj)
return;
#if defined(BIDI_DEBUG) && BIDI_DEBUG > 1
kdDebug(6041) << "appendRun: dir="<<(int)dir<<endl;
#endif
bool b = bidi.adjustEmbedding;
bidi.adjustEmbedding = false;
int start = bidi.sor.pos;
RenderObject *obj = bidi.sor.obj;
while (obj && obj != bidi.eor.obj && obj != bidi.endOfLine.obj) {
appendRunsForObject(start, obj->length(), obj, bidi);
start = 0;
obj = bidiNext(bidi.sor.block, obj, bidi);
}
if (obj) {
unsigned pos = obj == bidi.eor.obj ? bidi.eor.pos : UINT_MAX;
if (obj == bidi.endOfLine.obj && bidi.endOfLine.pos <= pos) {
bidi.reachedEndOfLine = true;
pos = bidi.endOfLine.pos;
}
// It's OK to add runs for zero-length RenderObjects, just don't make the run larger than it should be
int end = obj->length() ? pos+1 : 0;
appendRunsForObject(start, end, obj, bidi);
}
bidi.eor.increment(bidi);
bidi.sor = bidi.eor;
bidi.dir = WTF::Unicode::OtherNeutral;
bidi.status.eor = WTF::Unicode::OtherNeutral;
bidi.adjustEmbedding = b;
}
static void embed(WTF::Unicode::Direction d, BidiState& bidi)
{
bool b = bidi.adjustEmbedding;
bidi.adjustEmbedding = false;
if (d == WTF::Unicode::PopDirectionalFormat) {
BidiContext *c = bidi.context->parent;
if (c) {
if (!emptyRun && bidi.eor != bidi.last) {
assert(bidi.status.eor != WTF::Unicode::OtherNeutral);
// bidi.sor ... bidi.eor ... bidi.last eor; need to append the bidi.sor-bidi.eor run or extend it through bidi.last
assert(bidi.status.last == WTF::Unicode::EuropeanNumberSeparator
|| bidi.status.last == WTF::Unicode::EuropeanNumberTerminator
|| bidi.status.last == WTF::Unicode::CommonNumberSeparator
|| bidi.status.last == WTF::Unicode::BoundaryNeutral
|| bidi.status.last == WTF::Unicode::BlockSeparator
|| bidi.status.last == WTF::Unicode::SegmentSeparator
|| bidi.status.last == WTF::Unicode::WhiteSpaceNeutral
|| bidi.status.last == WTF::Unicode::OtherNeutral);
if (bidi.dir == WTF::Unicode::OtherNeutral)
bidi.dir = bidi.context->dir();
if (bidi.context->dir() == WTF::Unicode::LeftToRight) {
// bidi.sor ... bidi.eor ... bidi.last L
if (bidi.status.eor == WTF::Unicode::EuropeanNumber) {
if (bidi.status.lastStrong != WTF::Unicode::LeftToRight) {
bidi.dir = WTF::Unicode::EuropeanNumber;
appendRun(bidi);
}
} else if (bidi.status.eor == WTF::Unicode::ArabicNumber) {
bidi.dir = WTF::Unicode::ArabicNumber;
appendRun(bidi);
} else if (bidi.status.eor != WTF::Unicode::LeftToRight)
appendRun(bidi);
} else if (bidi.status.eor != WTF::Unicode::RightToLeft && bidi.status.eor != WTF::Unicode::RightToLeftArabic)
appendRun(bidi);
bidi.eor = bidi.last;
}
appendRun(bidi);
emptyRun = true;
// sor for the new run is determined by the higher level (rule X10)
bidi.status.last = bidi.context->dir();
bidi.status.lastStrong = bidi.context->dir();
bidi.context = c;
bidi.status.eor = bidi.context->dir();
bidi.eor.obj = 0;
}
} else {
WTF::Unicode::Direction runDir;
if (d == WTF::Unicode::RightToLeftEmbedding || d == WTF::Unicode::RightToLeftOverride)
runDir = WTF::Unicode::RightToLeft;
else
runDir = WTF::Unicode::LeftToRight;
bool override = d == WTF::Unicode::LeftToRightOverride || d == WTF::Unicode::RightToLeftOverride;
unsigned char level = bidi.context->level;
if (runDir == WTF::Unicode::RightToLeft) {
if (level%2) // we have an odd level
level += 2;
else
level++;
} else {
if (level%2) // we have an odd level
level++;
else
level += 2;
}
if (level < 61) {
if (!emptyRun && bidi.eor != bidi.last) {
assert(bidi.status.eor != WTF::Unicode::OtherNeutral);
// bidi.sor ... bidi.eor ... bidi.last eor; need to append the bidi.sor-bidi.eor run or extend it through bidi.last
assert(bidi.status.last == WTF::Unicode::EuropeanNumberSeparator
|| bidi.status.last == WTF::Unicode::EuropeanNumberTerminator
|| bidi.status.last == WTF::Unicode::CommonNumberSeparator
|| bidi.status.last == WTF::Unicode::BoundaryNeutral
|| bidi.status.last == WTF::Unicode::BlockSeparator
|| bidi.status.last == WTF::Unicode::SegmentSeparator
|| bidi.status.last == WTF::Unicode::WhiteSpaceNeutral
|| bidi.status.last == WTF::Unicode::OtherNeutral);
if (bidi.dir == WTF::Unicode::OtherNeutral)
bidi.dir = runDir;
if (runDir == WTF::Unicode::LeftToRight) {
// bidi.sor ... bidi.eor ... bidi.last L
if (bidi.status.eor == WTF::Unicode::EuropeanNumber) {
if (bidi.status.lastStrong != WTF::Unicode::LeftToRight) {
bidi.dir = WTF::Unicode::EuropeanNumber;
appendRun(bidi);
if (bidi.context->dir() != WTF::Unicode::LeftToRight)
bidi.dir = WTF::Unicode::RightToLeft;
}
} else if (bidi.status.eor == WTF::Unicode::ArabicNumber) {
bidi.dir = WTF::Unicode::ArabicNumber;
appendRun(bidi);
if (bidi.context->dir() != WTF::Unicode::LeftToRight) {
bidi.eor = bidi.last;
bidi.dir = WTF::Unicode::RightToLeft;
appendRun(bidi);
}
} else if (bidi.status.eor != WTF::Unicode::LeftToRight) {
if (bidi.context->dir() == WTF::Unicode::LeftToRight || bidi.status.lastStrong == WTF::Unicode::LeftToRight)
appendRun(bidi);
else
bidi.dir = WTF::Unicode::RightToLeft;
}
} else if (bidi.status.eor != WTF::Unicode::RightToLeft && bidi.status.eor != WTF::Unicode::RightToLeftArabic) {
// bidi.sor ... bidi.eor ... bidi.last R; bidi.eor=L/EN/AN; EN,AN behave like R (rule N1)
if (bidi.context->dir() == WTF::Unicode::RightToLeft || bidi.status.lastStrong == WTF::Unicode::RightToLeft || bidi.status.lastStrong == WTF::Unicode::RightToLeftArabic)
appendRun(bidi);
else
bidi.dir = WTF::Unicode::LeftToRight;
}
bidi.eor = bidi.last;
}
appendRun(bidi);
emptyRun = true;
bidi.context = new BidiContext(level, runDir, bidi.context.get(), override);
bidi.status.last = runDir;
bidi.status.lastStrong = runDir;
bidi.status.eor = runDir;
bidi.eor.obj = 0;
}
}
bidi.adjustEmbedding = b;
}
InlineFlowBox* RenderBlock::createLineBoxes(RenderObject* obj)
{
// See if we have an unconstructed line box for this object that is also
// the last item on the line.
ASSERT(obj->isInlineFlow() || obj == this);
RenderFlow* flow = static_cast<RenderFlow*>(obj);
// Get the last box we made for this render object.
InlineFlowBox* box = flow->lastLineBox();
// If this box is constructed then it is from a previous line, and we need
// to make a new box for our line. If this box is unconstructed but it has
// something following it on the line, then we know we have to make a new box
// as well. In this situation our inline has actually been split in two on
// the same line (this can happen with very fancy language mixtures).
if (!box || box->isConstructed() || box->nextOnLine()) {
// We need to make a new box for this render object. Once
// made, we need to place it at the end of the current line.
InlineBox* newBox = obj->createInlineBox(false, obj == this);
ASSERT(newBox->isInlineFlowBox());
box = static_cast<InlineFlowBox*>(newBox);
box->setFirstLineStyleBit(m_firstLine);
// We have a new box. Append it to the inline box we get by constructing our
// parent. If we have hit the block itself, then |box| represents the root
// inline box for the line, and it doesn't have to be appended to any parent
// inline.
if (obj != this) {
InlineFlowBox* parentBox = createLineBoxes(obj->parent());
parentBox->addToLine(box);
}
}
return box;
}
RootInlineBox* RenderBlock::constructLine(const BidiIterator& start, const BidiIterator& end)
{
if (!sFirstBidiRun)
return 0; // We had no runs. Don't make a root inline box at all. The line is empty.
InlineFlowBox* parentBox = 0;
for (BidiRun* r = sFirstBidiRun; r; r = r->nextRun) {
// Create a box for our object.
bool isOnlyRun = (sBidiRunCount == 1);
if (sBidiRunCount == 2 && !r->obj->isListMarker())
isOnlyRun = ((style()->direction() == RTL) ? sLastBidiRun : sFirstBidiRun)->obj->isListMarker();
r->box = r->obj->createInlineBox(r->obj->isPositioned(), false, isOnlyRun);
if (r->box) {
// If we have no parent box yet, or if the run is not simply a sibling,
// then we need to construct inline boxes as necessary to properly enclose the
// run's inline box.
if (!parentBox || parentBox->object() != r->obj->parent())
// Create new inline boxes all the way back to the appropriate insertion point.
parentBox = createLineBoxes(r->obj->parent());
// Append the inline box to this line.
parentBox->addToLine(r->box);
if (r->box->isInlineTextBox()) {
InlineTextBox *text = static_cast<InlineTextBox*>(r->box);
text->setStart(r->start);
text->setLen(r->stop - r->start);
bool visuallyOrdered = r->obj->style()->visuallyOrdered();
text->m_reversed = r->level % 2 && !visuallyOrdered;
text->m_dirOverride = r->override || visuallyOrdered;
}
}
}
// We should have a root inline box. It should be unconstructed and
// be the last continuation of our line list.
assert(lastLineBox() && !lastLineBox()->isConstructed());
// Set bits on our inline flow boxes that indicate which sides should
// paint borders/margins/padding. This knowledge will ultimately be used when
// we determine the horizontal positions and widths of all the inline boxes on
// the line.
RenderObject* endObject = 0;
bool lastLine = !end.obj;
if (end.obj && end.pos == 0)
endObject = end.obj;
lastLineBox()->determineSpacingForFlowBoxes(lastLine, endObject);
// Now mark the line boxes as being constructed.
lastLineBox()->setConstructed();
// Return the last line.
return lastRootBox();
}
// usage: tw - (xpos % tw);
int RenderBlock::tabWidth(bool isWhitespacePre)
{
if (!isWhitespacePre)
return 0;
if (m_tabWidth == -1) {
const UChar spaceChar = ' ';
const Font& font = style()->font();
int spaceWidth = font.width(TextRun(&spaceChar, 1));
m_tabWidth = spaceWidth * 8;
}
return m_tabWidth;
}
void RenderBlock::computeHorizontalPositionsForLine(RootInlineBox* lineBox, BidiState& bidi)
{
// First determine our total width.
int availableWidth = lineWidth(m_height);
int totWidth = lineBox->getFlowSpacingWidth();
BidiRun* r = 0;
bool needsWordSpacing = false;
for (r = sFirstBidiRun; r; r = r->nextRun) {
if (!r->box || r->obj->isPositioned() || r->box->isLineBreak())
continue; // Positioned objects are only participating to figure out their
// correct static x position. They have no effect on the width.
// Similarly, line break boxes have no effect on the width.
if (r->obj->isText()) {
RenderText* rt = static_cast<RenderText*>(r->obj);
int textWidth = rt->width(r->start, r->stop-r->start, totWidth, m_firstLine);
int effectiveWidth = textWidth;
int rtLength = rt->length();
if (rtLength != 0) {
if (r->start == 0 && needsWordSpacing && DeprecatedChar(rt->text()[r->start]).isSpace())
effectiveWidth += rt->font(m_firstLine)->wordSpacing();
needsWordSpacing = !DeprecatedChar(rt->text()[r->stop-1]).isSpace() && r->stop == rtLength;
}
r->box->setWidth(textWidth);
} else if (!r->obj->isInlineFlow()) {
r->obj->calcWidth();
r->box->setWidth(r->obj->width());
if (!r->compact)
totWidth += r->obj->marginLeft() + r->obj->marginRight();
}
// Compacts don't contribute to the width of the line, since they are placed in the margin.
if (!r->compact)
totWidth += r->box->width();
}
if (totWidth > availableWidth && sLogicallyLastBidiRun->obj->style(m_firstLine)->autoWrap() &&
sLogicallyLastBidiRun->obj->style(m_firstLine)->breakOnlyAfterWhiteSpace() &&
!sLogicallyLastBidiRun->compact) {
sLogicallyLastBidiRun->box->setWidth(sLogicallyLastBidiRun->box->width() - totWidth + availableWidth);
totWidth = availableWidth;
}
// Armed with the total width of the line (without justification),
// we now examine our text-align property in order to determine where to position the
// objects horizontally. The total width of the line can be increased if we end up
// justifying text.
int x = leftOffset(m_height);
switch(style()->textAlign()) {
case LEFT:
case KHTML_LEFT:
// The direction of the block should determine what happens with wide lines. In
// particular with RTL blocks, wide lines should still spill out to the left.
if (style()->direction() == RTL && totWidth > availableWidth)
x -= (totWidth - availableWidth);
numSpaces = 0;
break;
case JUSTIFY:
if (numSpaces != 0 && !bidi.current.atEnd() && !lineBox->endsWithBreak())
break;
// fall through
case TAAUTO:
numSpaces = 0;
// for right to left fall through to right aligned
if (bidi.context->basicDir() == WTF::Unicode::LeftToRight)
break;
case RIGHT:
case KHTML_RIGHT:
// Wide lines spill out of the block based off direction.
// So even if text-align is right, if direction is LTR, wide lines should overflow out of the right
// side of the block.
if (style()->direction() == RTL || totWidth < availableWidth)
x += availableWidth - totWidth;
numSpaces = 0;
break;
case CENTER:
case KHTML_CENTER:
int xd = (availableWidth - totWidth)/2;
x += xd > 0 ? xd : 0;
numSpaces = 0;
break;
}
if (numSpaces > 0) {
for (r = sFirstBidiRun; r; r = r->nextRun) {
if (!r->box) continue;
int spaceAdd = 0;
if (numSpaces > 0 && r->obj->isText() && !r->compact) {
// get the number of spaces in the run
int spaces = 0;
for ( int i = r->start; i < r->stop; i++ ) {
UChar c = static_cast<RenderText*>(r->obj)->text()[i];
if (c == ' ' || c == '\n' || c == '\t')
spaces++;
}
ASSERT(spaces <= numSpaces);
// Only justify text if whitespace is collapsed.
if (r->obj->style()->collapseWhiteSpace()) {
spaceAdd = (availableWidth - totWidth)*spaces/numSpaces;
static_cast<InlineTextBox*>(r->box)->setSpaceAdd(spaceAdd);
totWidth += spaceAdd;
}
numSpaces -= spaces;
}
}
}
// The widths of all runs are now known. We can now place every inline box (and
// compute accurate widths for the inline flow boxes).
int leftPosition = x;
int rightPosition = x;
needsWordSpacing = false;
lineBox->placeBoxesHorizontally(x, leftPosition, rightPosition, needsWordSpacing);
lineBox->setHorizontalOverflowPositions(leftPosition, rightPosition);
}
void RenderBlock::computeVerticalPositionsForLine(RootInlineBox* lineBox)
{
lineBox->verticallyAlignBoxes(m_height);
lineBox->setBlockHeight(m_height);
// See if the line spilled out. If so set overflow height accordingly.
int bottomOfLine = lineBox->bottomOverflow();
if (bottomOfLine > m_height && bottomOfLine > m_overflowHeight)
m_overflowHeight = bottomOfLine;
// Now make sure we place replaced render objects correctly.
for (BidiRun* r = sFirstBidiRun; r; r = r->nextRun) {
if (!r->box)
continue; // Skip runs with no line boxes.
// Align positioned boxes with the top of the line box. This is
// a reasonable approximation of an appropriate y position.
if (r->obj->isPositioned())
r->box->setYPos(m_height);
// Position is used to properly position both replaced elements and
// to update the static normal flow x/y of positioned elements.
r->obj->position(r->box);
}
}
// collects one line of the paragraph and transforms it to visual order
void RenderBlock::bidiReorderLine(const BidiIterator& start, const BidiIterator& end, BidiState& bidi)
{
if (start == end) {
if (start.current() == '\n')
m_height += lineHeight(m_firstLine, true);
return;
}
sFirstBidiRun = 0;
sLastBidiRun = 0;
sBidiRunCount = 0;
assert(bidi.dir == WTF::Unicode::OtherNeutral);
emptyRun = true;
bidi.eor.obj = 0;
numSpaces = 0;
bidi.current = start;
bidi.last = bidi.current;
bool pastEnd = false;
BidiState stateAtEnd;
while (true) {
WTF::Unicode::Direction dirCurrent;
if (pastEnd && (previousLineBrokeCleanly || bidi.current.atEnd())) {
BidiContext *c = bidi.context.get();
while (c->parent)
c = c->parent;
dirCurrent = c->dir();
if (previousLineBrokeCleanly) {
// A deviation from the Unicode Bidi Algorithm in order to match
// Mac OS X text and WinIE: a hard line break resets bidi state.
stateAtEnd.context = c;
stateAtEnd.status.eor = dirCurrent;
stateAtEnd.status.last = dirCurrent;
stateAtEnd.status.lastStrong = dirCurrent;
}
} else {
dirCurrent = bidi.current.direction();
if (bidi.context->override
&& dirCurrent != WTF::Unicode::RightToLeftEmbedding
&& dirCurrent != WTF::Unicode::LeftToRightEmbedding
&& dirCurrent != WTF::Unicode::RightToLeftOverride
&& dirCurrent != WTF::Unicode::LeftToRightOverride
&& dirCurrent != WTF::Unicode::PopDirectionalFormat)
dirCurrent = bidi.context->dir();
else if (dirCurrent == WTF::Unicode::NonSpacingMark)
dirCurrent = bidi.status.last;
}
assert(bidi.status.eor != WTF::Unicode::OtherNeutral);
switch (dirCurrent) {
// embedding and overrides (X1-X9 in the Bidi specs)
case WTF::Unicode::RightToLeftEmbedding:
case WTF::Unicode::LeftToRightEmbedding:
case WTF::Unicode::RightToLeftOverride:
case WTF::Unicode::LeftToRightOverride:
case WTF::Unicode::PopDirectionalFormat:
embed(dirCurrent, bidi);
break;
// strong types
case WTF::Unicode::LeftToRight:
switch(bidi.status.last) {
case WTF::Unicode::RightToLeft:
case WTF::Unicode::RightToLeftArabic:
case WTF::Unicode::EuropeanNumber:
case WTF::Unicode::ArabicNumber:
if (bidi.status.last != WTF::Unicode::EuropeanNumber || bidi.status.lastStrong != WTF::Unicode::LeftToRight)
appendRun(bidi);
break;
case WTF::Unicode::LeftToRight:
break;
case WTF::Unicode::EuropeanNumberSeparator:
case WTF::Unicode::EuropeanNumberTerminator:
case WTF::Unicode::CommonNumberSeparator:
case WTF::Unicode::BoundaryNeutral:
case WTF::Unicode::BlockSeparator:
case WTF::Unicode::SegmentSeparator:
case WTF::Unicode::WhiteSpaceNeutral:
case WTF::Unicode::OtherNeutral:
if (bidi.status.eor == WTF::Unicode::EuropeanNumber) {
if (bidi.status.lastStrong != WTF::Unicode::LeftToRight) {
// the numbers need to be on a higher embedding level, so let's close that run
bidi.dir = WTF::Unicode::EuropeanNumber;
appendRun(bidi);
if (bidi.context->dir() != WTF::Unicode::LeftToRight) {
// the neutrals take the embedding direction, which is R
bidi.eor = bidi.last;
bidi.dir = WTF::Unicode::RightToLeft;
appendRun(bidi);
}
}
} else if (bidi.status.eor == WTF::Unicode::ArabicNumber) {
// Arabic numbers are always on a higher embedding level, so let's close that run
bidi.dir = WTF::Unicode::ArabicNumber;
appendRun(bidi);
if (bidi.context->dir() != WTF::Unicode::LeftToRight) {
// the neutrals take the embedding direction, which is R
bidi.eor = bidi.last;
bidi.dir = WTF::Unicode::RightToLeft;
appendRun(bidi);
}
} else if(bidi.status.eor != WTF::Unicode::LeftToRight) {
//last stuff takes embedding dir
if (bidi.context->dir() != WTF::Unicode::LeftToRight && bidi.status.lastStrong != WTF::Unicode::LeftToRight) {
bidi.eor = bidi.last;
bidi.dir = WTF::Unicode::RightToLeft;
}
appendRun(bidi);
}
default:
break;
}
bidi.eor = bidi.current;
bidi.status.eor = WTF::Unicode::LeftToRight;
bidi.status.lastStrong = WTF::Unicode::LeftToRight;
bidi.dir = WTF::Unicode::LeftToRight;
break;
case WTF::Unicode::RightToLeftArabic:
case WTF::Unicode::RightToLeft:
switch (bidi.status.last) {
case WTF::Unicode::LeftToRight:
case WTF::Unicode::EuropeanNumber:
case WTF::Unicode::ArabicNumber:
appendRun(bidi);
case WTF::Unicode::RightToLeft:
case WTF::Unicode::RightToLeftArabic:
break;
case WTF::Unicode::EuropeanNumberSeparator:
case WTF::Unicode::EuropeanNumberTerminator:
case WTF::Unicode::CommonNumberSeparator:
case WTF::Unicode::BoundaryNeutral:
case WTF::Unicode::BlockSeparator:
case WTF::Unicode::SegmentSeparator:
case WTF::Unicode::WhiteSpaceNeutral:
case WTF::Unicode::OtherNeutral:
if (bidi.status.eor != WTF::Unicode::RightToLeft && bidi.status.eor != WTF::Unicode::RightToLeftArabic) {
//last stuff takes embedding dir
if (bidi.context->dir() != WTF::Unicode::RightToLeft && bidi.status.lastStrong != WTF::Unicode::RightToLeft
&& bidi.status.lastStrong != WTF::Unicode::RightToLeftArabic) {
bidi.eor = bidi.last;
bidi.dir = WTF::Unicode::LeftToRight;
}
appendRun(bidi);
}
default:
break;
}
bidi.eor = bidi.current;
bidi.status.eor = WTF::Unicode::RightToLeft;
bidi.status.lastStrong = dirCurrent;
bidi.dir = WTF::Unicode::RightToLeft;
break;
// weak types:
case WTF::Unicode::EuropeanNumber:
if (bidi.status.lastStrong != WTF::Unicode::RightToLeftArabic) {
// if last strong was AL change EN to AN
switch (bidi.status.last) {
case WTF::Unicode::EuropeanNumber:
case WTF::Unicode::LeftToRight:
break;
case WTF::Unicode::RightToLeft:
case WTF::Unicode::RightToLeftArabic:
case WTF::Unicode::ArabicNumber:
bidi.eor = bidi.last;
appendRun(bidi);
bidi.dir = WTF::Unicode::EuropeanNumber;
break;
case WTF::Unicode::EuropeanNumberSeparator:
case WTF::Unicode::CommonNumberSeparator:
if (bidi.status.eor == WTF::Unicode::EuropeanNumber)
break;
case WTF::Unicode::EuropeanNumberTerminator:
case WTF::Unicode::BoundaryNeutral:
case WTF::Unicode::BlockSeparator:
case WTF::Unicode::SegmentSeparator:
case WTF::Unicode::WhiteSpaceNeutral:
case WTF::Unicode::OtherNeutral:
if (bidi.status.eor == WTF::Unicode::RightToLeft) {
// neutrals go to R
bidi.eor = bidi.status.last == WTF::Unicode::EuropeanNumberTerminator ? bidi.lastBeforeET : bidi.last;
appendRun(bidi);
bidi.dir = WTF::Unicode::EuropeanNumber;
} else if (bidi.status.eor != WTF::Unicode::LeftToRight &&
(bidi.status.eor != WTF::Unicode::EuropeanNumber || bidi.status.lastStrong != WTF::Unicode::LeftToRight) &&
bidi.dir != WTF::Unicode::LeftToRight) {
// numbers on both sides, neutrals get right to left direction
appendRun(bidi);
bidi.eor = bidi.status.last == WTF::Unicode::EuropeanNumberTerminator ? bidi.lastBeforeET : bidi.last;
bidi.dir = WTF::Unicode::RightToLeft;
appendRun(bidi);
bidi.dir = WTF::Unicode::EuropeanNumber;
}
default:
break;
}
bidi.eor = bidi.current;
bidi.status.eor = WTF::Unicode::EuropeanNumber;
if (bidi.dir == WTF::Unicode::OtherNeutral)
bidi.dir = WTF::Unicode::LeftToRight;
break;
}
case WTF::Unicode::ArabicNumber:
dirCurrent = WTF::Unicode::ArabicNumber;
switch (bidi.status.last) {
case WTF::Unicode::LeftToRight:
if (bidi.context->dir() == WTF::Unicode::LeftToRight)
appendRun(bidi);
break;
case WTF::Unicode::ArabicNumber:
break;
case WTF::Unicode::RightToLeft:
case WTF::Unicode::RightToLeftArabic:
case WTF::Unicode::EuropeanNumber:
bidi.eor = bidi.last;
appendRun(bidi);
break;
case WTF::Unicode::CommonNumberSeparator:
if (bidi.status.eor == WTF::Unicode::ArabicNumber)
break;
case WTF::Unicode::EuropeanNumberSeparator:
case WTF::Unicode::EuropeanNumberTerminator:
case WTF::Unicode::BoundaryNeutral:
case WTF::Unicode::BlockSeparator:
case WTF::Unicode::SegmentSeparator:
case WTF::Unicode::WhiteSpaceNeutral:
case WTF::Unicode::OtherNeutral:
if (bidi.status.eor != WTF::Unicode::RightToLeft && bidi.status.eor != WTF::Unicode::RightToLeftArabic) {
// run of L before neutrals, neutrals take embedding dir (N2)
if (bidi.context->dir() == WTF::Unicode::RightToLeft || bidi.status.lastStrong == WTF::Unicode::RightToLeft
|| bidi.status.lastStrong == WTF::Unicode::RightToLeftArabic) {
// the embedding direction is R
// close the L run
appendRun(bidi);
// neutrals become an R run
bidi.dir = WTF::Unicode::RightToLeft;
} else {
// the embedding direction is L
// append neutrals to the L run and close it
bidi.dir = WTF::Unicode::LeftToRight;
}
}
bidi.eor = bidi.last;
appendRun(bidi);
default:
break;
}
bidi.eor = bidi.current;
bidi.status.eor = WTF::Unicode::ArabicNumber;
if (bidi.dir == WTF::Unicode::OtherNeutral)
bidi.dir = WTF::Unicode::ArabicNumber;
break;
case WTF::Unicode::EuropeanNumberSeparator:
case WTF::Unicode::CommonNumberSeparator:
break;
case WTF::Unicode::EuropeanNumberTerminator:
if (bidi.status.last == WTF::Unicode::EuropeanNumber) {
dirCurrent = WTF::Unicode::EuropeanNumber;
bidi.eor = bidi.current;
bidi.status.eor = dirCurrent;
} else if (bidi.status.last != WTF::Unicode::EuropeanNumberTerminator)
bidi.lastBeforeET = emptyRun ? bidi.eor : bidi.last;
break;
// boundary neutrals should be ignored
case WTF::Unicode::BoundaryNeutral:
if (bidi.eor == bidi.last)
bidi.eor = bidi.current;
break;
// neutrals
case WTF::Unicode::BlockSeparator:
// ### what do we do with newline and paragraph seperators that come to here?
break;
case WTF::Unicode::SegmentSeparator:
// ### implement rule L1
break;
case WTF::Unicode::WhiteSpaceNeutral:
break;
case WTF::Unicode::OtherNeutral:
break;
default:
break;
}
if (pastEnd) {
if (bidi.eor == bidi.current) {
if (!bidi.reachedEndOfLine) {
bidi.eor = bidi.endOfLine;
switch (bidi.status.eor) {
case WTF::Unicode::LeftToRight:
case WTF::Unicode::RightToLeft:
case WTF::Unicode::ArabicNumber:
bidi.dir = bidi.status.eor;
break;
case WTF::Unicode::EuropeanNumber:
bidi.dir = bidi.status.lastStrong == WTF::Unicode::LeftToRight ? WTF::Unicode::LeftToRight : WTF::Unicode::EuropeanNumber;
break;
default:
assert(false);
}
appendRun(bidi);
}
bidi = stateAtEnd;
bidi.dir = WTF::Unicode::OtherNeutral;
break;
}
}
// set status.last as needed.
switch (dirCurrent) {
case WTF::Unicode::EuropeanNumberTerminator:
if (bidi.status.last != WTF::Unicode::EuropeanNumber)
bidi.status.last = WTF::Unicode::EuropeanNumberTerminator;
break;
case WTF::Unicode::EuropeanNumberSeparator:
case WTF::Unicode::CommonNumberSeparator:
case WTF::Unicode::SegmentSeparator:
case WTF::Unicode::WhiteSpaceNeutral:
case WTF::Unicode::OtherNeutral:
switch(bidi.status.last) {
case WTF::Unicode::LeftToRight:
case WTF::Unicode::RightToLeft:
case WTF::Unicode::RightToLeftArabic:
case WTF::Unicode::EuropeanNumber:
case WTF::Unicode::ArabicNumber:
bidi.status.last = dirCurrent;
break;
default:
bidi.status.last = WTF::Unicode::OtherNeutral;
}
break;
case WTF::Unicode::NonSpacingMark:
case WTF::Unicode::BoundaryNeutral:
case WTF::Unicode::RightToLeftEmbedding:
case WTF::Unicode::LeftToRightEmbedding:
case WTF::Unicode::RightToLeftOverride:
case WTF::Unicode::LeftToRightOverride:
case WTF::Unicode::PopDirectionalFormat:
// ignore these
break;
case WTF::Unicode::EuropeanNumber:
// fall through
default:
bidi.status.last = dirCurrent;
}
bidi.last = bidi.current;
if (emptyRun && !(dirCurrent == WTF::Unicode::RightToLeftEmbedding
|| dirCurrent == WTF::Unicode::LeftToRightEmbedding
|| dirCurrent == WTF::Unicode::RightToLeftOverride
|| dirCurrent == WTF::Unicode::LeftToRightOverride
|| dirCurrent == WTF::Unicode::PopDirectionalFormat)) {
bidi.sor = bidi.current;
emptyRun = false;
}
// this causes the operator ++ to open and close embedding levels as needed
// for the CSS unicode-bidi property
bidi.adjustEmbedding = true;
bidi.current.increment(bidi);
bidi.adjustEmbedding = false;
if (emptyRun && (dirCurrent == WTF::Unicode::RightToLeftEmbedding
|| dirCurrent == WTF::Unicode::LeftToRightEmbedding
|| dirCurrent == WTF::Unicode::RightToLeftOverride
|| dirCurrent == WTF::Unicode::LeftToRightOverride
|| dirCurrent == WTF::Unicode::PopDirectionalFormat)) {
// exclude the embedding char itself from the new run so that ATSUI will never see it
bidi.eor.obj = 0;
bidi.last = bidi.current;
bidi.sor = bidi.current;
}
if (!pastEnd && (bidi.current == end || bidi.current.atEnd())) {
if (emptyRun)
break;
stateAtEnd = bidi;
bidi.endOfLine = bidi.last;
pastEnd = true;
}
}
sLogicallyLastBidiRun = sLastBidiRun;
// reorder line according to run structure...
// do not reverse for visually ordered web sites
if (!style()->visuallyOrdered()) {
// first find highest and lowest levels
unsigned char levelLow = 128;
unsigned char levelHigh = 0;
BidiRun* r = sFirstBidiRun;
while (r) {
if (r->level > levelHigh)
levelHigh = r->level;
if (r->level < levelLow)
levelLow = r->level;
r = r->nextRun;
}
// implements reordering of the line (L2 according to Bidi spec):
// L2. From the highest level found in the text to the lowest odd level on each line,
// reverse any contiguous sequence of characters that are at that level or higher.
// reversing is only done up to the lowest odd level
if (!(levelLow%2))
levelLow++;
int count = sBidiRunCount - 1;
while (levelHigh >= levelLow) {
int i = 0;
BidiRun* currRun = sFirstBidiRun;
while (i < count) {
while (i < count && currRun && currRun->level < levelHigh) {
i++;
currRun = currRun->nextRun;
}
int start = i;
while (i <= count && currRun && currRun->level >= levelHigh) {
i++;
currRun = currRun->nextRun;
}
int end = i-1;
reverseRuns(start, end);
}
levelHigh--;
}
}
bidi.endOfLine.obj = 0;
}
static void buildCompactRuns(RenderObject* compactObj, BidiState& bidi)
{
sBuildingCompactRuns = true;
if (!compactObj->isRenderBlock()) {
// Just append a run for our object.
isLineEmpty = false;
addRun(new (compactObj->renderArena()) BidiRun(0, compactObj->length(), compactObj, bidi.context.get(), bidi.dir));
}
else {
// Format the compact like it is its own single line. We build up all the runs for
// the little compact and then reorder them for bidi.
RenderBlock* compactBlock = static_cast<RenderBlock*>(compactObj);
bidi.adjustEmbedding = true;
BidiIterator start(compactBlock, bidiFirst(compactBlock, bidi), 0);
bidi.adjustEmbedding = false;
BidiIterator end = start;
betweenMidpoints = false;
isLineEmpty = true;
previousLineBrokeCleanly = true;
end = compactBlock->findNextLineBreak(start, bidi);
if (!isLineEmpty)
compactBlock->bidiReorderLine(start, end, bidi);
}
sCompactFirstBidiRun = sFirstBidiRun;
sCompactLastBidiRun = sLastBidiRun;
sCompactBidiRunCount = sBidiRunCount;
sNumMidpoints = 0;
sCurrMidpoint = 0;
betweenMidpoints = false;
sBuildingCompactRuns = false;
}
IntRect RenderBlock::layoutInlineChildren(bool relayoutChildren)
{
BidiState bidi;
bool useRepaintRect = false;
int repaintTop = 0;
int repaintBottom = 0;
m_overflowHeight = 0;
invalidateVerticalPositions();
m_height = borderTop() + paddingTop();
int toAdd = borderBottom() + paddingBottom();
if (includeHorizontalScrollbarSize())
toAdd += m_layer->horizontalScrollbarHeight();
// Figure out if we should clear out our line boxes.
// FIXME: Handle resize eventually!
// FIXME: Do something better when floats are present.
bool fullLayout = !firstLineBox() || !firstChild() || selfNeedsLayout() || relayoutChildren || containsFloats();
if (fullLayout)
deleteLineBoxes();
// Text truncation only kicks in if your overflow isn't visible and your text-overflow-mode isn't
// clip.
// FIXME: CSS3 says that descendants that are clipped must also know how to truncate. This is insanely
// difficult to figure out (especially in the middle of doing layout), and is really an esoteric pile of nonsense
// anyway, so we won't worry about following the draft here.
bool hasTextOverflow = style()->textOverflow() && hasOverflowClip();
// Walk all the lines and delete our ellipsis line boxes if they exist.
if (hasTextOverflow)
deleteEllipsisLineBoxes();
if (firstChild()) {
// layout replaced elements
bool endOfInline = false;
RenderObject *o = bidiFirst(this, bidi, false);
bool hasFloat = false;
while (o) {
if (o->isReplaced() || o->isFloating() || o->isPositioned()) {
if (relayoutChildren || o->style()->width().isPercent() || o->style()->height().isPercent())
o->setChildNeedsLayout(true, false);
if (o->isPositioned())
o->containingBlock()->insertPositionedObject(o);
else {
if (o->isFloating())
hasFloat = true;
else if (fullLayout || o->needsLayout()) // Replaced elements
o->dirtyLineBoxes(fullLayout);
o->layoutIfNeeded();
}
}
else if (o->isText() || (o->isInlineFlow() && !endOfInline)) {
if (fullLayout || o->selfNeedsLayout())
o->dirtyLineBoxes(fullLayout);
o->setNeedsLayout(false);
}
o = bidiNext(this, o, bidi, false, &endOfInline);
}
if (hasFloat)
fullLayout = true; // FIXME: Will need to find a way to optimize floats some day.
if (fullLayout && !selfNeedsLayout()) {
setNeedsLayout(true, false); // Mark ourselves as needing a full layout. This way we'll repaint like
// we're supposed to.
if (!document()->view()->needsFullRepaint() && m_layer) {
// Because we waited until we were already inside layout to discover
// that the block really needed a full layout, we missed our chance to repaint the layer
// before layout started. Luckily the layer has cached the repaint rect for its original
// position and size, and so we can use that to make a repaint happen now.
RenderView* c = view();
if (c && !c->printingMode())
c->repaintViewRectangle(m_layer->repaintRect());
}
}
BidiContext *startEmbed;
if (style()->direction() == LTR) {
startEmbed = new BidiContext( 0, WTF::Unicode::LeftToRight, NULL, style()->unicodeBidi() == Override );
bidi.status.eor = WTF::Unicode::LeftToRight;
} else {
startEmbed = new BidiContext( 1, WTF::Unicode::RightToLeft, NULL, style()->unicodeBidi() == Override );
bidi.status.eor = WTF::Unicode::RightToLeft;
}
bidi.status.lastStrong = startEmbed->dir();
bidi.status.last = startEmbed->dir();
bidi.status.eor = startEmbed->dir();
bidi.context = startEmbed;
bidi.dir = WTF::Unicode::OtherNeutral;
if (!smidpoints)
smidpoints = new Vector<BidiIterator>();
sNumMidpoints = 0;
sCurrMidpoint = 0;
sCompactFirstBidiRun = sCompactLastBidiRun = 0;
sCompactBidiRunCount = 0;
// We want to skip ahead to the first dirty line
BidiIterator start;
RootInlineBox* startLine = determineStartPosition(fullLayout, start, bidi);
// We also find the first clean line and extract these lines. We will add them back
// if we determine that we're able to synchronize after handling all our dirty lines.
BidiIterator cleanLineStart;
BidiStatus cleanLineBidiStatus;
BidiContext* cleanLineBidiContext;
int endLineYPos;
RootInlineBox* endLine = (fullLayout || !startLine) ?
0 : determineEndPosition(startLine, cleanLineStart, cleanLineBidiStatus, cleanLineBidiContext, endLineYPos);
if (endLine && cleanLineBidiContext)
cleanLineBidiContext->ref();
if (startLine) {
useRepaintRect = true;
repaintTop = m_height;
repaintBottom = m_height;
RenderArena* arena = renderArena();
RootInlineBox* box = startLine;
while (box) {
repaintTop = min(repaintTop, box->topOverflow());
repaintBottom = max(repaintBottom, box->bottomOverflow());
RootInlineBox* next = box->nextRootBox();
box->deleteLine(arena);
box = next;
}
startLine = 0;
}
BidiIterator end = start;
bool endLineMatched = false;
while (!end.atEnd()) {
start = end;
if (endLine && (endLineMatched = matchedEndLine(start, bidi.status, bidi.context.get(), cleanLineStart, cleanLineBidiStatus, cleanLineBidiContext, endLine, endLineYPos, repaintBottom, repaintTop)))
break;
betweenMidpoints = false;
isLineEmpty = true;
if (m_firstLine && firstChild() && firstChild()->isCompact() && firstChild()->isRenderBlock()) {
buildCompactRuns(firstChild(), bidi);
start.obj = firstChild()->nextSibling();
end = start;
}
end = findNextLineBreak(start, bidi);
if (start.atEnd())
break;
if (!isLineEmpty) {
bidiReorderLine(start, end, bidi);
// Now that the runs have been ordered, we create the line boxes.
// At the same time we figure out where border/padding/margin should be applied for
// inline flow boxes.
if (sCompactFirstBidiRun) {
// We have a compact line sharing this line. Link the compact runs
// to our runs to create a single line of runs.
sCompactLastBidiRun->nextRun = sFirstBidiRun;
sFirstBidiRun = sCompactFirstBidiRun;
sBidiRunCount += sCompactBidiRunCount;
}
RootInlineBox* lineBox = 0;
if (sBidiRunCount) {
lineBox = constructLine(start, end);
if (lineBox) {
lineBox->setEndsWithBreak(previousLineBrokeCleanly);
// Now we position all of our text runs horizontally.
computeHorizontalPositionsForLine(lineBox, bidi);
// Now position our text runs vertically.
computeVerticalPositionsForLine(lineBox);
#if PLATFORM(MAC)
// Highlight acts as an overflow inflation.
if (style()->highlight() != nullAtom)
lineBox->addHighlightOverflow();
#endif
deleteBidiRuns(renderArena());
}
}
if (end == start) {
bidi.adjustEmbedding = true;
end.increment(bidi);
bidi.adjustEmbedding = false;
}
if (lineBox) {
lineBox->setLineBreakInfo(end.obj, end.pos, &bidi.status, bidi.context.get());
if (useRepaintRect) {
repaintTop = min(repaintTop, lineBox->topOverflow());
repaintBottom = max(repaintBottom, lineBox->bottomOverflow());
}
}
m_firstLine = false;
newLine();
}
sNumMidpoints = 0;
sCurrMidpoint = 0;
sCompactFirstBidiRun = sCompactLastBidiRun = 0;
sCompactBidiRunCount = 0;
}
if (endLine) {
if (endLineMatched) {
// Attach all the remaining lines, and then adjust their y-positions as needed.
for (RootInlineBox* line = endLine; line; line = line->nextRootBox())
line->attachLine();
// Now apply the offset to each line if needed.
int delta = m_height - endLineYPos;
if (delta) {
for (RootInlineBox* line = endLine; line; line = line->nextRootBox()) {
repaintTop = min(repaintTop, line->topOverflow() + (delta < 0 ? delta : 0));
repaintBottom = max(repaintBottom, line->bottomOverflow() + (delta > 0 ? delta : 0));
line->adjustPosition(0, delta);
}
}
m_height = lastRootBox()->blockHeight();
} else {
// Delete all the remaining lines.
InlineRunBox* line = endLine;
RenderArena* arena = renderArena();
while (line) {
repaintTop = min(repaintTop, line->topOverflow());
repaintBottom = max(repaintBottom, line->bottomOverflow());
InlineRunBox* next = line->nextLineBox();
line->deleteLine(arena);
line = next;
}
}
if (cleanLineBidiContext)
cleanLineBidiContext->deref();
}
}
sNumMidpoints = 0;
sCurrMidpoint = 0;
// in case we have a float on the last line, it might not be positioned up to now.
// This has to be done before adding in the bottom border/padding, or the float will
// include the padding incorrectly. -dwh
positionNewFloats();
// Now add in the bottom border/padding.
m_height += toAdd;
// Always make sure this is at least our height.
m_overflowHeight = max(m_height, m_overflowHeight);
// See if any lines spill out of the block. If so, we need to update our overflow width.
checkLinesForOverflow();
IntRect repaintRect(0, 0, 0, 0);
if (useRepaintRect) {
repaintRect.setX(m_overflowLeft);
repaintRect.setWidth(m_overflowWidth - m_overflowLeft);
repaintRect.setY(repaintTop);
repaintRect.setHeight(repaintBottom - repaintTop);
}
if (!firstLineBox() && hasLineIfEmpty())
m_height += lineHeight(true, true);
// See if we have any lines that spill out of our block. If we do, then we will possibly need to
// truncate text.
if (hasTextOverflow)
checkLinesForTextOverflow();
return repaintRect;
#if defined(BIDI_DEBUG) && BIDI_DEBUG > 1
kdDebug(6041) << " ------- bidi end " << this << " -------" << endl;
#endif
}
RootInlineBox* RenderBlock::determineStartPosition(bool fullLayout, BidiIterator& start, BidiState& bidi)
{
RootInlineBox* curr = 0;
RootInlineBox* last = 0;
RenderObject* startObj = 0;
int pos = 0;
if (fullLayout) {
// Nuke all our lines.
if (firstRootBox()) {
RenderArena* arena = renderArena();
curr = firstRootBox();
while (curr) {
RootInlineBox* next = curr->nextRootBox();
curr->deleteLine(arena);
curr = next;
}
ASSERT(!m_firstLineBox && !m_lastLineBox);
}
} else {
for (curr = firstRootBox(); curr && !curr->isDirty(); curr = curr->nextRootBox());
if (curr) {
// We have a dirty line.
if (curr->prevRootBox()) {
// We have a previous line.
if (!curr->prevRootBox()->endsWithBreak())
curr = curr->prevRootBox(); // The previous line didn't break cleanly, so treat it as dirty also.
}
} else {
// No dirty lines were found.
// If the last line didn't break cleanly, treat it as dirty.
if (lastRootBox() && !lastRootBox()->endsWithBreak())
curr = lastRootBox();
}
// If we have no dirty lines, then last is just the last root box.
last = curr ? curr->prevRootBox() : lastRootBox();
}
m_firstLine = !last;
previousLineBrokeCleanly = !last || last->endsWithBreak();
if (last) {
m_height = last->blockHeight();
startObj = last->lineBreakObj();
pos = last->lineBreakPos();
bidi.status = last->lineBreakBidiStatus();
bidi.context = last->lineBreakBidiContext();
} else {
bidi.adjustEmbedding = true;
startObj = bidiFirst(this, bidi, 0);
bidi.adjustEmbedding = false;
}
start = BidiIterator(this, startObj, pos);
return curr;
}
RootInlineBox* RenderBlock::determineEndPosition(RootInlineBox* startLine, BidiIterator& cleanLineStart,
BidiStatus& cleanLineBidiStatus, BidiContext*& cleanLineBidiContext,
int& yPos)
{
RootInlineBox* last = 0;
if (!startLine)
last = 0;
else {
for (RootInlineBox* curr = startLine->nextRootBox(); curr; curr = curr->nextRootBox()) {
if (curr->isDirty())
last = 0;
else if (!last)
last = curr;
}
}
if (!last)
return 0;
RootInlineBox* prev = last->prevRootBox();
cleanLineStart = BidiIterator(this, prev->lineBreakObj(), prev->lineBreakPos());
cleanLineBidiStatus = prev->lineBreakBidiStatus();
cleanLineBidiContext = prev->lineBreakBidiContext();
yPos = prev->blockHeight();
for (RootInlineBox* line = last; line; line = line->nextRootBox())
line->extractLine(); // Disconnect all line boxes from their render objects while preserving
// their connections to one another.
return last;
}
bool RenderBlock::matchedEndLine(const BidiIterator& start, const BidiStatus& status, BidiContext* context,
const BidiIterator& endLineStart, const BidiStatus& endLineStatus, BidiContext* endLineContext,
RootInlineBox*& endLine, int& endYPos, int& repaintBottom, int& repaintTop)
{
if (start == endLineStart)
return status == endLineStatus && *context == *endLineContext;
else {
// The first clean line doesn't match, but we can check a handful of following lines to try
// to match back up.
static int numLines = 8; // The # of lines we're willing to match against.
RootInlineBox* line = endLine;
for (int i = 0; i < numLines && line; i++, line = line->nextRootBox()) {
if (line->lineBreakObj() == start.obj && line->lineBreakPos() == start.pos) {
// We have a match.
if (line->lineBreakBidiStatus() != status || *line->lineBreakBidiContext() != *context)
return false; // ...but the bidi state doesn't match.
RootInlineBox* result = line->nextRootBox();
// Set our yPos to be the block height of endLine.
if (result)
endYPos = line->blockHeight();
// Now delete the lines that we failed to sync.
RootInlineBox* boxToDelete = endLine;
RenderArena* arena = renderArena();
while (boxToDelete && boxToDelete != result) {
repaintTop = min(repaintTop, boxToDelete->topOverflow());
repaintBottom = max(repaintBottom, boxToDelete->bottomOverflow());
RootInlineBox* next = boxToDelete->nextRootBox();
boxToDelete->deleteLine(arena);
boxToDelete = next;
}
endLine = result;
return result;
}
}
}
return false;
}
static const unsigned short nonBreakingSpace = 0xa0;
static inline bool skipNonBreakingSpace(BidiIterator &it)
{
if (it.obj->style()->nbspMode() != SPACE || it.current() != nonBreakingSpace)
return false;
// FIXME: This is bad. It makes nbsp inconsistent with space and won't work correctly
// with m_minWidth/m_maxWidth.
// Do not skip a non-breaking space if it is the first character
// on a line after a clean line break (or on the first line, since previousLineBrokeCleanly starts off
// |true|).
if (isLineEmpty && previousLineBrokeCleanly)
return false;
return true;
}
static inline bool shouldCollapseWhiteSpace(const RenderStyle* style)
{
return style->collapseWhiteSpace() || (style->whiteSpace() == PRE_WRAP && (!isLineEmpty || !previousLineBrokeCleanly));
}
int RenderBlock::skipWhitespace(BidiIterator &it, BidiState &bidi)
{
// FIXME: The entire concept of the skipWhitespace function is flawed, since we really need to be building
// line boxes even for containers that may ultimately collapse away. Otherwise we'll never get positioned
// elements quite right. In other words, we need to build this function's work into the normal line
// object iteration process.
int w = lineWidth(m_height);
while (!it.atEnd() && (it.obj->isInlineFlow() || (shouldCollapseWhiteSpace(it.obj->style()) && !it.obj->isBR() &&
(it.current() == ' ' || it.current() == '\t' || (!it.obj->style()->preserveNewline() && it.current() == '\n') ||
it.current() == SOFT_HYPHEN || skipNonBreakingSpace(it) || it.obj->isFloatingOrPositioned())))) {
if (it.obj->isFloatingOrPositioned()) {
RenderObject *o = it.obj;
// add to special objects...
if (o->isFloating()) {
insertFloatingObject(o);
positionNewFloats();
w = lineWidth(m_height);
}
else if (o->isPositioned()) {
// FIXME: The math here is actually not really right. It's a best-guess approximation that
// will work for the common cases
RenderObject* c = o->container();
if (c->isInlineFlow()) {
// A relative positioned inline encloses us. In this case, we also have to determine our
// position as though we were an inline. Set |staticX| and |staticY| on the relative positioned
// inline so that we can obtain the value later.
c->setStaticX(style()->direction() == LTR ?
leftOffset(m_height) : rightOffset(m_height));
c->setStaticY(m_height);
}
if (o->hasStaticX()) {
bool wasInline = o->style()->isOriginalDisplayInlineType();
if (wasInline)
o->setStaticX(style()->direction() == LTR ?
leftOffset(m_height) :
width() - rightOffset(m_height));
else
o->setStaticX(style()->direction() == LTR ?
borderLeft() + paddingLeft() :
borderRight() + paddingRight());
}
if (o->hasStaticY())
o->setStaticY(m_height);
}
}
bidi.adjustEmbedding = true;
it.increment(bidi);
bidi.adjustEmbedding = false;
}
return w;
}
BidiIterator RenderBlock::findNextLineBreak(BidiIterator &start, BidiState &bidi)
{
// eliminate spaces at beginning of line
int width = skipWhitespace(start, bidi);
int w = 0;
int tmpW = 0;
if (start.atEnd())
return start;
// This variable is used only if whitespace isn't set to PRE, and it tells us whether
// or not we are currently ignoring whitespace.
bool ignoringSpaces = false;
BidiIterator ignoreStart;
// This variable tracks whether the very last character we saw was a space. We use
// this to detect when we encounter a second space so we know we have to terminate
// a run.
bool currentCharacterIsSpace = false;
bool currentCharacterIsWS = false;
RenderObject* trailingSpaceObject = 0;
BidiIterator lBreak = start;
RenderObject *o = start.obj;
RenderObject *last = o;
RenderObject *previous = o;
int pos = start.pos;
bool prevLineBrokeCleanly = previousLineBrokeCleanly;
previousLineBrokeCleanly = false;
while (o) {
if (o->isBR()) {
if (w + tmpW <= width) {
lBreak.obj = o;
lBreak.pos = 0;
lBreak.increment(bidi);
// A <br> always breaks a line, so don't let the line be collapsed
// away. Also, the space at the end of a line with a <br> does not
// get collapsed away. It only does this if the previous line broke
// cleanly. Otherwise the <br> has no effect on whether the line is
// empty or not.
if (prevLineBrokeCleanly)
isLineEmpty = false;
trailingSpaceObject = 0;
previousLineBrokeCleanly = true;
if (!isLineEmpty) {
// only check the clear status for non-empty lines.
EClear clear = o->style()->clear();
if (clear != CNONE)
m_clearStatus = (EClear) (m_clearStatus | clear);
}
}
goto end;
}
if (o->isFloatingOrPositioned()) {
// add to special objects...
if (o->isFloating()) {
insertFloatingObject(o);
// check if it fits in the current line.
// If it does, position it now, otherwise, position
// it after moving to next line (in newLine() func)
if (o->width()+o->marginLeft()+o->marginRight()+w+tmpW <= width) {
positionNewFloats();
width = lineWidth(m_height);
}
} else if (o->isPositioned()) {
// If our original display wasn't an inline type, then we can
// go ahead and determine our static x position now.
bool isInlineType = o->style()->isOriginalDisplayInlineType();
bool needToSetStaticX = o->hasStaticX();
if (o->hasStaticX() && !isInlineType) {
o->setStaticX(o->parent()->style()->direction() == LTR ?
borderLeft()+paddingLeft() :
borderRight()+paddingRight());
needToSetStaticX = false;
}
// If our original display was an INLINE type, then we can go ahead
// and determine our static y position now.
bool needToSetStaticY = o->hasStaticY();
if (o->hasStaticY() && isInlineType) {
o->setStaticY(m_height);
needToSetStaticY = false;
}
bool needToCreateLineBox = needToSetStaticX || needToSetStaticY;
RenderObject* c = o->container();
if (c->isInlineFlow() && (!needToSetStaticX || !needToSetStaticY))
needToCreateLineBox = true;
// If we're ignoring spaces, we have to stop and include this object and
// then start ignoring spaces again.
if (needToCreateLineBox) {
trailingSpaceObject = 0;
ignoreStart.obj = o;
ignoreStart.pos = 0;
if (ignoringSpaces) {
addMidpoint(ignoreStart); // Stop ignoring spaces.
addMidpoint(ignoreStart); // Start ignoring again.
}
}
}
} else if (o->isInlineFlow()) {
// Only empty inlines matter. We treat those similarly to replaced elements.
assert(!o->firstChild());
tmpW += o->marginLeft()+o->borderLeft()+o->paddingLeft()+
o->marginRight()+o->borderRight()+o->paddingRight();
} else if (o->isReplaced()) {
EWhiteSpace currWS = o->style()->whiteSpace();
EWhiteSpace lastWS = last->style()->whiteSpace();
// WinIE marquees have different whitespace characteristics by default when viewed from
// the outside vs. the inside. Text inside is NOWRAP, and so we altered the marquee's
// style to reflect this, but we now have to get back to the original whitespace value
// for the marquee when checking for line breaking.
if (o->isHTMLMarquee() && o->layer() && o->layer()->marquee())
currWS = o->layer()->marquee()->whiteSpace();
if (last->isHTMLMarquee() && last->layer() && last->layer()->marquee())
lastWS = last->layer()->marquee()->whiteSpace();
// Break on replaced elements if either has normal white-space.
// FIXME: This does not match WinIE, Opera, and Mozilla. They treat replaced elements
// like characters in a word, and require spaces between the replaced elements in order
// to break.
if (RenderStyle::autoWrap(currWS) || RenderStyle::autoWrap(lastWS)) {
w += tmpW;
tmpW = 0;
lBreak.obj = o;
lBreak.pos = 0;
}
if (ignoringSpaces) {
BidiIterator startMid( 0, o, 0 );
addMidpoint(startMid);
}
isLineEmpty = false;
ignoringSpaces = false;
currentCharacterIsSpace = false;
currentCharacterIsWS = false;
trailingSpaceObject = 0;
if (o->isListMarker() && o->style()->listStylePosition() == OUTSIDE) {
// The marker must not have an effect on whitespace at the start
// of the line. We start ignoring spaces to make sure that any additional
// spaces we see will be discarded.
//
// Optimize for a common case. If we can't find whitespace after the list
// item, then this is all moot. -dwh
RenderObject* next = bidiNext(start.block, o, bidi);
if (style()->collapseWhiteSpace() && next && !next->isBR() && next->isText() && static_cast<RenderText*>(next)->stringLength() > 0) {
RenderText *nextText = static_cast<RenderText*>(next);
UChar nextChar = nextText->text()[0];
if (nextText->style()->isCollapsibleWhiteSpace(nextChar)) {
currentCharacterIsSpace = true;
currentCharacterIsWS = true;
ignoringSpaces = true;
BidiIterator endMid( 0, o, 0 );
addMidpoint(endMid);
}
}
} else
tmpW += o->width()+o->marginLeft()+o->marginRight()+inlineWidth(o);
} else if (o->isText()) {
RenderText *t = static_cast<RenderText *>(o);
int strlen = t->stringLength();
int len = strlen - pos;
const UChar* str = t->text();
const Font *f = t->font(m_firstLine);
// proportional font, needs a bit more work.
int lastSpace = pos;
int wordSpacing = o->style()->wordSpacing();
int lastSpaceWordSpacing = 0;
bool appliedStartWidth = pos > 0; // If the span originated on a previous line,
// then assume the start width has been applied.
int wrapW = tmpW + inlineWidth(o, !appliedStartWidth, true);
int nextBreakable = -1;
while (len) {
bool previousCharacterIsSpace = currentCharacterIsSpace;
bool previousCharacterIsWS = currentCharacterIsWS;
UChar c = str[pos];
currentCharacterIsSpace = c == ' ' || c == '\t' || (!o->style()->preserveNewline() && (c == '\n'));
if (!o->style()->collapseWhiteSpace() || !currentCharacterIsSpace)
isLineEmpty = false;
// Check for soft hyphens. Go ahead and ignore them.
if (c == SOFT_HYPHEN) {
if (!ignoringSpaces) {
// Ignore soft hyphens
BidiIterator endMid;
if (pos > 0)
endMid = BidiIterator(0, o, pos-1);
else
endMid = BidiIterator(0, previous, previous->isText() ? static_cast<RenderText *>(previous)->stringLength() - 1 : 0);
// Two consecutive soft hyphens. Avoid overlapping midpoints.
if (sNumMidpoints && smidpoints->at(sNumMidpoints - 1).obj == endMid.obj && smidpoints->at(sNumMidpoints - 1).pos > endMid.pos)
sNumMidpoints--;
else
addMidpoint(endMid);
// Add the width up to but not including the hyphen.
tmpW += t->width(lastSpace, pos - lastSpace, f, w+tmpW) + lastSpaceWordSpacing;
// For wrapping text only, include the hyphen. We need to ensure it will fit
// on the line if it shows when we break.
if (o->style()->autoWrap())
tmpW += t->width(pos, 1, f, w+tmpW);
BidiIterator startMid(0, o, pos+1);
addMidpoint(startMid);
}
pos++;
len--;
lastSpaceWordSpacing = 0;
lastSpace = pos; // Cheesy hack to prevent adding in widths of the run twice.
continue;
}
bool applyWordSpacing = false;
bool allowBreak = o->style()->autoWrap();
bool breakNBSP = allowBreak && o->style()->nbspMode() == SPACE;
// FIXME: This check looks suspicious. Why does w have to be 0?
bool breakWords = o->style()->wordWrap() == BREAK_WORD && ((allowBreak && w == 0) || o->style()->whiteSpace() == PRE);
currentCharacterIsWS = currentCharacterIsSpace || (breakNBSP && c == nonBreakingSpace);
if (breakWords)
wrapW += t->width(pos, 1, f, w+wrapW);
bool midWordBreak = breakWords && (w + wrapW > width);
if (c == '\n' || (o->style()->whiteSpace() != PRE && isBreakable(str, pos, strlen, nextBreakable, breakNBSP)) || midWordBreak) {
bool stoppedIgnoringSpaces = false;
if (ignoringSpaces) {
if (!currentCharacterIsSpace) {
// Stop ignoring spaces and begin at this
// new point.
ignoringSpaces = false;
lastSpaceWordSpacing = 0;
lastSpace = pos; // e.g., "Foo goo", don't add in any of the ignored spaces.
BidiIterator startMid ( 0, o, pos );
addMidpoint(startMid);
stoppedIgnoringSpaces = true;
} else {
// Just keep ignoring these spaces.
pos++;
len--;
continue;
}
}
int additionalTmpW = t->width(lastSpace, pos - lastSpace, f, w+tmpW) + lastSpaceWordSpacing;
tmpW += additionalTmpW;
if (!appliedStartWidth) {
tmpW += inlineWidth(o, true, false);
appliedStartWidth = true;
}
applyWordSpacing = wordSpacing && currentCharacterIsSpace && !previousCharacterIsSpace;
if (o->style()->autoWrap() && w + tmpW > width && w == 0) {
int fb = nearestFloatBottom(m_height);
int newLineWidth = lineWidth(fb);
// See if |tmpW| will fit on the new line. As long as it does not,
// keep adjusting our float bottom until we find some room.
int lastFloatBottom = m_height;
while (lastFloatBottom < fb && tmpW > newLineWidth) {
lastFloatBottom = fb;
fb = nearestFloatBottom(fb);
newLineWidth = lineWidth(fb);
}
if (!w && m_height < fb && width < newLineWidth) {
m_height = fb;
width = newLineWidth;
}
}
if (o->style()->autoWrap() || breakWords) {
// If we break only after white-space, consider the current character
// as candidate width for this line.
bool lineWasTooWide = false;
if (w + tmpW <= width && currentCharacterIsWS && o->style()->breakOnlyAfterWhiteSpace() && !midWordBreak) {
int charWidth = t->width(pos, 1, f, w + tmpW) + (applyWordSpacing ? wordSpacing : 0);
// Check if line is too big even without the extra space
// at the end of the line. If it is not, do nothing.
// If the line needs the extra whitespace to be too long,
// then move the line break to the space and skip all
// additional whitespace.
if (w + tmpW + charWidth > width) {
lineWasTooWide = true;
lBreak.obj = o;
lBreak.pos = pos;
if (pos > 0) {
// Separate the trailing space into its own box, which we will
// resize to fit on the line in computeHorizontalPositionsForLine().
BidiIterator midpoint(0, o, pos);
addMidpoint(BidiIterator(0, o, pos-1)); // Stop
addMidpoint(BidiIterator(0, o, pos)); // Start
}
skipWhitespace(lBreak, bidi);
}
}
if (lineWasTooWide || w + tmpW > width) {
if (lBreak.obj && lBreak.obj->style()->preserveNewline() && lBreak.obj->isText() && static_cast<RenderText*>(lBreak.obj)->text()[lBreak.pos] == '\n') {
if (!stoppedIgnoringSpaces && pos > 0) {
// We need to stop right before the newline and then start up again.
BidiIterator midpoint(0, o, pos);
addMidpoint(BidiIterator(0, o, pos-1)); // Stop
addMidpoint(BidiIterator(0, o, pos)); // Start
}
lBreak.increment(bidi);
previousLineBrokeCleanly = true;
}
goto end; // Didn't fit. Jump to the end.
} else {
if (midWordBreak)
tmpW -= additionalTmpW;
if (pos > 0 && str[pos-1] == SOFT_HYPHEN)
// Subtract the width of the soft hyphen out since we fit on a line.
tmpW -= t->width(pos-1, 1, f, w+tmpW);
}
}
if (c == '\n' && o->style()->preserveNewline()) {
if (!stoppedIgnoringSpaces && pos > 0) {
// We need to stop right before the newline and then start up again.
BidiIterator midpoint(0, o, pos);
addMidpoint(BidiIterator(0, o, pos-1)); // Stop
addMidpoint(BidiIterator(0, o, pos)); // Start
}
lBreak.obj = o;
lBreak.pos = pos;
lBreak.increment(bidi);
previousLineBrokeCleanly = true;
return lBreak;
}
if (o->style()->autoWrap()) {
w += tmpW;
tmpW = 0;
lBreak.obj = o;
lBreak.pos = pos;
}
if (midWordBreak) {
// Remember this as a breakable position in case
// adding the end width forces a break.
lBreak.obj = o;
lBreak.pos = pos;
} else {
lastSpaceWordSpacing = applyWordSpacing ? wordSpacing : 0;
lastSpace = pos;
}
if (!ignoringSpaces && o->style()->collapseWhiteSpace()) {
// If we encounter a newline, or if we encounter a
// second space, we need to go ahead and break up this
// run and enter a mode where we start collapsing spaces.
if (currentCharacterIsSpace && previousCharacterIsSpace) {
ignoringSpaces = true;
// We just entered a mode where we are ignoring
// spaces. Create a midpoint to terminate the run
// before the second space.
addMidpoint(ignoreStart);
}
}
} else if (ignoringSpaces) {
// Stop ignoring spaces and begin at this
// new point.
ignoringSpaces = false;
lastSpaceWordSpacing = applyWordSpacing ? wordSpacing : 0;
lastSpace = pos; // e.g., "Foo goo", don't add in any of the ignored spaces.
BidiIterator startMid(0, o, pos);
addMidpoint(startMid);
}
if (currentCharacterIsSpace && !previousCharacterIsSpace) {
ignoreStart.obj = o;
ignoreStart.pos = pos;
}
if (!currentCharacterIsWS && previousCharacterIsWS) {
if (o->style()->autoWrap() && o->style()->breakOnlyAfterWhiteSpace()) {
lBreak.obj = o;
lBreak.pos = pos;
}
}
if (o->style()->collapseWhiteSpace() && currentCharacterIsSpace && !ignoringSpaces)
trailingSpaceObject = o;
else if (!o->style()->collapseWhiteSpace() || !currentCharacterIsSpace)
trailingSpaceObject = 0;
pos++;
len--;
}
// IMPORTANT: pos is > length here!
if (!ignoringSpaces)
tmpW += t->width(lastSpace, pos - lastSpace, f, w+tmpW) + lastSpaceWordSpacing;
if (!appliedStartWidth)
tmpW += inlineWidth(o, true, false);
tmpW += inlineWidth(o, false, true);
} else
ASSERT( false );
RenderObject* next = bidiNext(start.block, o, bidi);
bool autoWrap = o->style()->autoWrap();
bool checkForBreak = autoWrap;
if (w && w + tmpW > width && lBreak.obj && o->style()->whiteSpace() == NOWRAP)
checkForBreak = true;
else if (next && o->isText() && next->isText() && !next->isBR()) {
if (autoWrap || (next->style()->autoWrap())) {
if (currentCharacterIsSpace)
checkForBreak = true;
else {
checkForBreak = false;
RenderText* nextText = static_cast<RenderText*>(next);
if (nextText->stringLength() != 0) {
UChar c = nextText->text()[0];
if (c == ' ' || c == '\t' || (c == '\n' && !next->style()->preserveNewline()))
// If the next item on the line is text, and if we did not end with
// a space, then the next text run continues our word (and so it needs to
// keep adding to |tmpW|. Just update and continue.
checkForBreak = true;
}
bool canPlaceOnLine = (w + tmpW <= width) || !autoWrap;
if (canPlaceOnLine && checkForBreak) {
w += tmpW;
tmpW = 0;
lBreak.obj = next;
lBreak.pos = 0;
}
}
}
}
if (checkForBreak && (w + tmpW > width)) {
// if we have floats, try to get below them.
if (currentCharacterIsSpace && !ignoringSpaces && o->style()->collapseWhiteSpace())
trailingSpaceObject = 0;
int fb = nearestFloatBottom(m_height);
int newLineWidth = lineWidth(fb);
// See if |tmpW| will fit on the new line. As long as it does not,
// keep adjusting our float bottom until we find some room.
int lastFloatBottom = m_height;
while (lastFloatBottom < fb && tmpW > newLineWidth) {
lastFloatBottom = fb;
fb = nearestFloatBottom(fb);
newLineWidth = lineWidth(fb);
}
if (!w && m_height < fb && width < newLineWidth) {
m_height = fb;
width = newLineWidth;
}
// |width| may have been adjusted because we got shoved down past a float (thus
// giving us more room), so we need to retest, and only jump to
// the end label if we still don't fit on the line. -dwh
if (w + tmpW > width)
goto end;
}
last = o;
if (!o->isFloating() && (!o->isPositioned() || o->hasStaticX() || o->hasStaticY() || !o->container()->isInlineFlow()))
previous = o;
o = next;
if (!last->isFloatingOrPositioned() && last->isReplaced() && last->style()->autoWrap() &&
(!last->isListMarker() || last->style()->listStylePosition()==INSIDE)) {
// Go ahead and add in tmpW.
w += tmpW;
tmpW = 0;
lBreak.obj = o;
lBreak.pos = 0;
}
// Clear out our character space bool, since inline <pre>s don't collapse whitespace
// with adjacent inline normal/nowrap spans.
if (!last->style()->collapseWhiteSpace())
currentCharacterIsSpace = false;
pos = 0;
}
if (w + tmpW <= width || (last && last->style()->whiteSpace() == NOWRAP)) {
lBreak.obj = 0;
lBreak.pos = 0;
}
end:
if (lBreak == start && !lBreak.obj->isBR()) {
// we just add as much as possible
if (style()->preserveNewline()) {
// FIXME: Don't really understand this case.
if (pos != 0) {
lBreak.obj = o;
lBreak.pos = pos - 1;
} else {
lBreak.obj = last;
lBreak.pos = last->isText() ? last->length() : 0;
}
} else if (lBreak.obj) {
if (last != o && !last->isListMarker()) {
// better to break between object boundaries than in the middle of a word (except for list markers)
lBreak.obj = o;
lBreak.pos = 0;
} else {
// Don't ever break in the middle of a word if we can help it.
// There's no room at all. We just have to be on this line,
// even though we'll spill out.
lBreak.obj = o;
lBreak.pos = pos;
}
}
}
// make sure we consume at least one char/object.
if (lBreak == start)
lBreak.increment(bidi);
// Sanity check our midpoints.
checkMidpoints(lBreak, bidi);
if (trailingSpaceObject) {
// This object is either going to be part of the last midpoint, or it is going
// to be the actual endpoint. In both cases we just decrease our pos by 1 level to
// exclude the space, allowing it to - in effect - collapse into the newline.
if (sNumMidpoints%2==1) {
BidiIterator* midpoints = smidpoints->data();
midpoints[sNumMidpoints-1].pos--;
}
//else if (lBreak.pos > 0)
// lBreak.pos--;
else if (lBreak.obj == 0 && trailingSpaceObject->isText()) {
// Add a new end midpoint that stops right at the very end.
RenderText* text = static_cast<RenderText *>(trailingSpaceObject);
unsigned pos = text->length() >=2 ? text->length() - 2 : UINT_MAX;
BidiIterator endMid(0, trailingSpaceObject, pos);
addMidpoint(endMid);
}
}
// We might have made lBreak an iterator that points past the end
// of the object. Do this adjustment to make it point to the start
// of the next object instead to avoid confusing the rest of the
// code.
if (lBreak.pos > 0) {
lBreak.pos--;
lBreak.increment(bidi);
}
if (lBreak.obj && lBreak.pos >= 2 && lBreak.obj->isText()) {
// For soft hyphens on line breaks, we have to chop out the midpoints that made us
// ignore the hyphen so that it will render at the end of the line.
UChar c = static_cast<RenderText*>(lBreak.obj)->text()[lBreak.pos-1];
if (c == SOFT_HYPHEN)
chopMidpointsAt(lBreak.obj, lBreak.pos-2);
}
return lBreak;
}
void RenderBlock::checkLinesForOverflow()
{
m_overflowWidth = m_width;
for (RootInlineBox* curr = firstRootBox(); curr; curr = curr->nextRootBox()) {
m_overflowLeft = min(curr->leftOverflow(), m_overflowLeft);
m_overflowTop = min(curr->topOverflow(), m_overflowTop);
m_overflowWidth = max(curr->rightOverflow(), m_overflowWidth);
m_overflowHeight = max(curr->bottomOverflow(), m_overflowHeight);
}
}
void RenderBlock::deleteEllipsisLineBoxes()
{
for (RootInlineBox* curr = firstRootBox(); curr; curr = curr->nextRootBox())
curr->clearTruncation();
}
void RenderBlock::checkLinesForTextOverflow()
{
// Determine the width of the ellipsis using the current font.
const UChar ellipsis = 0x2026; // FIXME: CSS3 says this is configurable, also need to use 0x002E (FULL STOP) if 0x2026 not renderable
TextRun ellipsisRun(&ellipsis, 1);
static AtomicString ellipsisStr(&ellipsis, 1);
const Font& firstLineFont = firstLineStyle()->font();
const Font& font = style()->font();
int firstLineEllipsisWidth = firstLineFont.width(ellipsisRun);
int ellipsisWidth = (font == firstLineFont) ? firstLineEllipsisWidth : font.width(ellipsisRun);
// For LTR text truncation, we want to get the right edge of our padding box, and then we want to see
// if the right edge of a line box exceeds that. For RTL, we use the left edge of the padding box and
// check the left edge of the line box to see if it is less
// Include the scrollbar for overflow blocks, which means we want to use "contentWidth()"
bool ltr = style()->direction() == LTR;
for (RootInlineBox* curr = firstRootBox(); curr; curr = curr->nextRootBox()) {
int blockEdge = ltr ? rightOffset(curr->yPos()) : leftOffset(curr->yPos());
int lineBoxEdge = ltr ? curr->xPos() + curr->width() : curr->xPos();
if ((ltr && lineBoxEdge > blockEdge) || (!ltr && lineBoxEdge < blockEdge)) {
// This line spills out of our box in the appropriate direction. Now we need to see if the line
// can be truncated. In order for truncation to be possible, the line must have sufficient space to
// accommodate our truncation string, and no replaced elements (images, tables) can overlap the ellipsis
// space.
int width = curr == firstRootBox() ? firstLineEllipsisWidth : ellipsisWidth;
if (curr->canAccommodateEllipsis(ltr, blockEdge, lineBoxEdge, width))
curr->placeEllipsis(ellipsisStr, ltr, blockEdge, width);
}
}
}
}