coretextcontroller.cpp

linux下开源浏览器WebKit的源码,市面上的很多商用浏览器都是移植自WebKit

/*
 * Copyright (C) 2007, 2008 Apple Inc. All Rights Reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 *
 * THIS SOFTWARE IS PROVIDED BY APPLE INC. ``AS IS'' AND ANY
 * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
 * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL APPLE INC. OR
 * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
 * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
 * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
 * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
 * OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 
 */

#include "config.h"
#include "CoreTextController.h"

#if USE(CORE_TEXT)

#include "CharacterNames.h"
#include "Font.h"
#include "FontCache.h"
#include "SimpleFontData.h"
#include "TextBreakIterator.h"
#include <wtf/MathExtras.h>

using namespace std;

namespace WebCore {

static inline CGFloat roundCGFloat(CGFloat f)
{
    if (sizeof(CGFloat) == sizeof(float))
        return roundf(static_cast<float>(f));
    return static_cast<CGFloat>(round(f));
}

static inline CGFloat ceilCGFloat(CGFloat f)
{
    if (sizeof(CGFloat) == sizeof(float))
        return ceilf(static_cast<float>(f));
    return static_cast<CGFloat>(ceil(f));
}

CoreTextController::CoreTextRun::CoreTextRun(CTRunRef ctRun, const SimpleFontData* fontData, const UChar* characters, unsigned stringLocation, size_t stringLength)
    : m_CTRun(ctRun)
    , m_fontData(fontData)
    , m_characters(characters)
    , m_stringLocation(stringLocation)
    , m_stringLength(stringLength)
{
    m_glyphCount = CTRunGetGlyphCount(ctRun);
    m_indices = CTRunGetStringIndicesPtr(ctRun);
    if (!m_indices) {
        m_indicesData.adoptCF(CFDataCreateMutable(kCFAllocatorDefault, m_glyphCount * sizeof(CFIndex)));
        CFDataIncreaseLength(m_indicesData.get(), m_glyphCount * sizeof(CFIndex));
        m_indices = reinterpret_cast<const CFIndex*>(CFDataGetMutableBytePtr(m_indicesData.get()));
        CTRunGetStringIndices(ctRun, CFRangeMake(0, 0), const_cast<CFIndex*>(m_indices));
    }
}

// Missing glyphs run constructor. Core Text will not generate a run of missing glyphs, instead falling back on
// glyphs from LastResort. We want to use the primary font's missing glyph in order to match the fast text code path.
CoreTextController::CoreTextRun::CoreTextRun(const SimpleFontData* fontData, const UChar* characters, unsigned stringLocation, size_t stringLength, bool ltr)
    : m_fontData(fontData)
    , m_characters(characters)
    , m_stringLocation(stringLocation)
    , m_stringLength(stringLength)
{
    Vector<CFIndex, 16> indices;
    unsigned r = 0;
    while (r < stringLength) {
        indices.append(r);
        if (U_IS_SURROGATE(characters[r])) {
            ASSERT(r + 1 < stringLength);
            ASSERT(U_IS_SURROGATE_LEAD(characters[r]));
            ASSERT(U_IS_TRAIL(characters[r + 1]));
            r += 2;
        } else
            r++;
    }
    m_glyphCount = indices.size();
    if (!ltr) {
        for (unsigned r = 0, end = m_glyphCount - 1; r < m_glyphCount / 2; ++r, --end)
            std::swap(indices[r], indices[end]);
    }
    m_indicesData.adoptCF(CFDataCreateMutable(kCFAllocatorDefault, m_glyphCount * sizeof(CFIndex)));
    CFDataAppendBytes(m_indicesData.get(), reinterpret_cast<const UInt8*>(indices.data()), m_glyphCount * sizeof(CFIndex));
    m_indices = reinterpret_cast<const CFIndex*>(CFDataGetBytePtr(m_indicesData.get()));
}

CoreTextController::CoreTextController(const Font* font, const TextRun& run, bool mayUseNaturalWritingDirection)
    : m_font(*font)
    , m_run(run)
    , m_mayUseNaturalWritingDirection(mayUseNaturalWritingDirection)
    , m_currentCharacter(0)
    , m_end(run.length())
    , m_totalWidth(0)
    , m_runWidthSoFar(0)
    , m_numGlyphsSoFar(0)
    , m_currentRun(0)
    , m_glyphInCurrentRun(0)
    , m_finalRoundingWidth(0)
    , m_lastRoundingGlyph(0)
{
    m_padding = m_run.padding();
    if (!m_padding)
        m_padPerSpace = 0;
    else {
        float numSpaces = 0;
        for (int s = 0; s < m_run.length(); s++)
            if (Font::treatAsSpace(m_run[s]))
                numSpaces++;

        if (numSpaces == 0)
            m_padPerSpace = 0;
        else
            m_padPerSpace = ceilf(m_run.padding() / numSpaces);
    }

    collectCoreTextRuns();
    adjustGlyphsAndAdvances();
}

int CoreTextController::offsetForPosition(int h, bool includePartialGlyphs)
{
    // FIXME: For positions occurring within a ligature, we should return the closest "ligature caret" or
    // approximate it by dividing the width of the ligature by the number of characters it encompasses.
    // However, Core Text does not expose a low-level API for directly finding
    // out how many characters a ligature encompasses (the "attachment count").
    if (h >= m_totalWidth)
        return m_run.ltr() ? m_end : 0;
    if (h < 0)
        return m_run.ltr() ? 0 : m_end;

    CGFloat x = h;

    size_t runCount = m_coreTextRuns.size();
    size_t offsetIntoAdjustedGlyphs = 0;

    for (size_t r = 0; r < runCount; ++r) {
        const CoreTextRun& coreTextRun = m_coreTextRuns[r];
        for (unsigned j = 0; j < coreTextRun.glyphCount(); ++j) {
            CGFloat adjustedAdvance = m_adjustedAdvances[offsetIntoAdjustedGlyphs + j].width;
            if (x <= adjustedAdvance) {
                CFIndex hitIndex = coreTextRun.indexAt(j);
                int stringLength = coreTextRun.stringLength();
                TextBreakIterator* characterIterator = characterBreakIterator(coreTextRun.characters(), stringLength);
                int clusterStart;
                if (isTextBreak(characterIterator, hitIndex))
                    clusterStart = hitIndex;
                else {
                    clusterStart = textBreakPreceding(characterIterator, hitIndex);
                    if (clusterStart == TextBreakDone)
                        clusterStart = 0;
                }

                if (!includePartialGlyphs)
                    return coreTextRun.stringLocation() + clusterStart;

                int clusterEnd = textBreakFollowing(characterIterator, hitIndex);
                if (clusterEnd == TextBreakDone)
                    clusterEnd = stringLength;

                CGFloat clusterWidth = adjustedAdvance;
                // FIXME: The search stops at the boundaries of coreTextRun. In theory, it should go on into neighboring CoreTextRuns
                // derived from the same CTLine. In practice, we do not expect there to be more than one CTRun in a CTLine, as no
                // reordering and on font fallback should occur within a CTLine.
                if (clusterEnd - clusterStart > 1) {
                    int firstGlyphBeforeCluster = j - 1;
                    while (firstGlyphBeforeCluster && coreTextRun.indexAt(firstGlyphBeforeCluster) >= clusterStart && coreTextRun.indexAt(firstGlyphBeforeCluster) < clusterEnd) {
                        CGFloat width = m_adjustedAdvances[offsetIntoAdjustedGlyphs + firstGlyphBeforeCluster].width;
                        clusterWidth += width;
                        x += width;
                        firstGlyphBeforeCluster--;
                    }
                    unsigned firstGlyphAfterCluster = j + 1;
                    while (firstGlyphAfterCluster < coreTextRun.glyphCount() && coreTextRun.indexAt(firstGlyphAfterCluster) >= clusterStart && coreTextRun.indexAt(firstGlyphAfterCluster) < clusterEnd) {
                        clusterWidth += m_adjustedAdvances[offsetIntoAdjustedGlyphs + firstGlyphAfterCluster].width;
                        firstGlyphAfterCluster++;
                    }
                }
                if (x <= clusterWidth / 2)
                    return coreTextRun.stringLocation() + (m_run.ltr() ? clusterStart : clusterEnd);
                else
                    return coreTextRun.stringLocation() + (m_run.ltr() ? clusterEnd : clusterStart);
            }
            x -= adjustedAdvance;
        }
        offsetIntoAdjustedGlyphs += coreTextRun.glyphCount();
    }

    ASSERT_NOT_REACHED();
    return 0;
}

void CoreTextController::collectCoreTextRuns()
{
    if (!m_end)
        return;

    // We break up glyph run generation for the string by FontData and (if needed) the use of small caps.
    const UChar* cp = m_run.characters();
    bool hasTrailingSoftHyphen = m_run[m_end - 1] == softHyphen;

    if (m_font.isSmallCaps() || hasTrailingSoftHyphen)
        m_smallCapsBuffer.resize(m_end);

    unsigned indexOfFontTransition = m_run.rtl() ? m_end - 1 : 0;
    const UChar* curr = m_run.rtl() ? cp + m_end  - 1 : cp;
    const UChar* end = m_run.rtl() ? cp - 1 : cp + m_end;

    // FIXME: Using HYPHEN-MINUS rather than HYPHEN because Times has a HYPHEN-MINUS glyph that looks like its
    // SOFT-HYPHEN glyph, and has no HYPHEN glyph.
    static const UChar hyphen = '-';

    if (hasTrailingSoftHyphen && m_run.rtl()) {
        collectCoreTextRunsForCharacters(&hyphen, 1, m_end - 1, m_font.glyphDataForCharacter(hyphen, false).fontData);
        indexOfFontTransition--;
        curr--;
    }

    GlyphData glyphData;
    GlyphData nextGlyphData;

    bool isSurrogate = U16_IS_SURROGATE(*curr);
    if (isSurrogate) {
        if (m_run.ltr()) {
            if (!U16_IS_SURROGATE_LEAD(curr[0]) || curr + 1 == end || !U16_IS_TRAIL(curr[1]))
                return;
            nextGlyphData = m_font.glyphDataForCharacter(U16_GET_SUPPLEMENTARY(curr[0], curr[1]), false);
        } else {
            if (!U16_IS_TRAIL(curr[0]) || curr -1 == end || !U16_IS_SURROGATE_LEAD(curr[-1]))
                return;
            nextGlyphData = m_font.glyphDataForCharacter(U16_GET_SUPPLEMENTARY(curr[-1], curr[0]), false);
        }
    } else
        nextGlyphData = m_font.glyphDataForCharacter(*curr, false);

    UChar newC = 0;

    bool isSmallCaps;
    bool nextIsSmallCaps = !isSurrogate && m_font.isSmallCaps() && !(U_GET_GC_MASK(*curr) & U_GC_M_MASK) && (newC = u_toupper(*curr)) != *curr;

    if (nextIsSmallCaps)
        m_smallCapsBuffer[curr - cp] = newC;

    while (true) {
        curr = m_run.rtl() ? curr - (isSurrogate ? 2 : 1) : curr + (isSurrogate ? 2 : 1);
        if (curr == end)
            break;

        glyphData = nextGlyphData;
        isSmallCaps = nextIsSmallCaps;
        int index = curr - cp;
        isSurrogate = U16_IS_SURROGATE(*curr);