widthiterator.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) 2003, 2006, 2008 Apple Inc. All rights reserved.
 * Copyright (C) 2008 Holger Hans Peter Freyther
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Library General Public
 * License as published by the Free Software Foundation; either
 * version 2 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Library General Public License for more details.
 *
 * You should have received a copy of the GNU Library General Public License
 * along with this library; see the file COPYING.LIB.  If not, write to
 * the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
 * Boston, MA 02110-1301, USA.
 *
 */

#include "config.h"
#include "WidthIterator.h"

#include "Font.h"
#include "GlyphBuffer.h"
#include "SimpleFontData.h"
#include <wtf/MathExtras.h>

#if USE(ICU_UNICODE)
#include <unicode/unorm.h>
#endif

using namespace WTF;
using namespace Unicode;

namespace WebCore {

// According to http://www.unicode.org/Public/UNIDATA/UCD.html#Canonical_Combining_Class_Values
static const uint8_t hiraganaKatakanaVoicingMarksCombiningClass = 8;

WidthIterator::WidthIterator(const Font* font, const TextRun& run)
    : m_font(font)
    , m_run(run)
    , m_end(run.length())
    , m_currentCharacter(0)
    , m_runWidthSoFar(0)
    , m_finalRoundingWidth(0)
{
    // If the padding is non-zero, count the number of spaces in the run
    // and divide that by the padding for per space addition.
    m_padding = m_run.padding();
    if (!m_padding)
        m_padPerSpace = 0;
    else {
        float numSpaces = 0;
        for (int i = 0; i < run.length(); i++)
            if (Font::treatAsSpace(m_run[i]))
                numSpaces++;

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

void WidthIterator::advance(int offset, GlyphBuffer* glyphBuffer)
{
    if (offset > m_end)
        offset = m_end;

    int currentCharacter = m_currentCharacter;
    const UChar* cp = m_run.data(currentCharacter);

    bool rtl = m_run.rtl();
    bool hasExtraSpacing = (m_font->letterSpacing() || m_font->wordSpacing() || m_padding) && !m_run.spacingDisabled();

    float runWidthSoFar = m_runWidthSoFar;
    float lastRoundingWidth = m_finalRoundingWidth;
    
    while (currentCharacter < offset) {
        UChar32 c = *cp;
        unsigned clusterLength = 1;
        if (c >= 0x3041) {
            if (c <= 0x30FE) {
                // Deal with Hiragana and Katakana voiced and semi-voiced syllables.
                // Normalize into composed form, and then look for glyph with base + combined mark.
                // Check above for character range to minimize performance impact.
                UChar32 normalized = normalizeVoicingMarks(currentCharacter);
                if (normalized) {
                    c = normalized;
                    clusterLength = 2;
                }
            } else if (U16_IS_SURROGATE(c)) {
                if (!U16_IS_SURROGATE_LEAD(c))
                    break;

                // Do we have a surrogate pair?  If so, determine the full Unicode (32 bit)
                // code point before glyph lookup.
                // Make sure we have another character and it's a low surrogate.
                if (currentCharacter + 1 >= m_run.length())
                    break;
                UChar low = cp[1];
                if (!U16_IS_TRAIL(low))
                    break;
                c = U16_GET_SUPPLEMENTARY(c, low);
                clusterLength = 2;
            }
        }

        const GlyphData& glyphData = m_font->glyphDataForCharacter(c, rtl);
        Glyph glyph = glyphData.glyph;
        const SimpleFontData* fontData = glyphData.fontData;

        ASSERT(fontData);

        // Now that we have a glyph and font data, get its width.
        float width;
        if (c == '\t' && m_run.allowTabs()) {
            float tabWidth = m_font->tabWidth();
            width = tabWidth - fmodf(m_run.xPos() + runWidthSoFar, tabWidth);
        } else {
            width = fontData->widthForGlyph(glyph);
            // We special case spaces in two ways when applying word rounding.
            // First, we round spaces to an adjusted width in all fonts.
            // Second, in fixed-pitch fonts we ensure that all characters that
            // match the width of the space character have the same width as the space character.
            if (width == fontData->m_spaceWidth && (fontData->m_treatAsFixedPitch || glyph == fontData->m_spaceGlyph) && m_run.applyWordRounding())
                width = fontData->m_adjustedSpaceWidth;
        }

        if (hasExtraSpacing) {
            // Account for letter-spacing.
            if (width && m_font->letterSpacing())
                width += m_font->letterSpacing();

            if (Font::treatAsSpace(c)) {
                // Account for padding. WebCore uses space padding to justify text.
                // We distribute the specified padding over the available spaces in the run.
                if (m_padding) {
                    // Use left over padding if not evenly divisible by number of spaces.
                    if (m_padding < m_padPerSpace) {
                        width += m_padding;
                        m_padding = 0;
                    } else {
                        width += m_padPerSpace;
                        m_padding -= m_padPerSpace;
                    }
                }

                // Account for word spacing.
                // We apply additional space between "words" by adding width to the space character.
                if (currentCharacter != 0 && !Font::treatAsSpace(cp[-1]) && m_font->wordSpacing())
                    width += m_font->wordSpacing();
            }
        }

        // Advance past the character we just dealt with.
        cp += clusterLength;
        currentCharacter += clusterLength;

        // Account for float/integer impedance mismatch between CG and KHTML. "Words" (characters 
        // followed by a character defined by isRoundingHackCharacter()) are always an integer width.
        // We adjust the width of the last character of a "word" to ensure an integer width.
        // If we move KHTML to floats we can remove this (and related) hacks.

        float oldWidth = width;

        // Force characters that are used to determine word boundaries for the rounding hack
        // to be integer width, so following words will start on an integer boundary.
        if (m_run.applyWordRounding() && Font::isRoundingHackCharacter(c))
            width = ceilf(width);

        // Check to see if the next character is a "rounding hack character", if so, adjust
        // width so that the total run width will be on an integer boundary.
        if ((m_run.applyWordRounding() && currentCharacter < m_run.length() && Font::isRoundingHackCharacter(*cp))
                || (m_run.applyRunRounding() && currentCharacter >= m_end)) {
            float totalWidth = runWidthSoFar + width;
            width += ceilf(totalWidth) - totalWidth;
        }

        runWidthSoFar += width;

        if (glyphBuffer)
            glyphBuffer->add(glyph, fontData, (rtl ? oldWidth + lastRoundingWidth : width));

        lastRoundingWidth = width - oldWidth;
    }

    m_currentCharacter = currentCharacter;
    m_runWidthSoFar = runWidthSoFar;
    m_finalRoundingWidth = lastRoundingWidth;
}

bool WidthIterator::advanceOneCharacter(float& width, GlyphBuffer* glyphBuffer)
{
    glyphBuffer->clear();
    advance(m_currentCharacter + 1, glyphBuffer);
    float w = 0;
    for (int i = 0; i < glyphBuffer->size(); ++i)
        w += glyphBuffer->advanceAt(i);
    width = w;
    return !glyphBuffer->isEmpty();
}

UChar32 WidthIterator::normalizeVoicingMarks(int currentCharacter)
{
    if (currentCharacter + 1 < m_end) {
        if (combiningClass(m_run[currentCharacter + 1]) == hiraganaKatakanaVoicingMarksCombiningClass) {
#if USE(ICU_UNICODE)
            // Normalize into composed form using 3.2 rules.
            UChar normalizedCharacters[2] = { 0, 0 };
            UErrorCode uStatus = U_ZERO_ERROR;  
            int32_t resultLength = unorm_normalize(m_run.data(currentCharacter), 2,
                UNORM_NFC, UNORM_UNICODE_3_2, &normalizedCharacters[0], 2, &uStatus);
            if (resultLength == 1 && uStatus == 0)
                return normalizedCharacters[0];
#elif USE(QT4_UNICODE)
            QString tmp(reinterpret_cast<const QChar*>(m_run.data(currentCharacter)), 2);
            QString res = tmp.normalized(QString::NormalizationForm_C, QChar::Unicode_3_2);
            if (res.length() == 1)
                return res.at(0).unicode();
#endif
        }
    }
    return 0;
}

}