qtexthtmlparser.cpp

奇趣公司比较新的qt/emd版本

    // find corresponding open node
    int p = last();
    if (p > 0
        && at(p - 1).tag == tag
        && at(p - 1).mayNotHaveChildren())
        p--;

    while (p && at(p).tag != tag)
        p = at(p).parent;

    // simply ignore the tag if we can't find
    // a corresponding open node, for broken
    // html such as <font>blah</font></font>
    if (!p)
        return;

    // in a white-space preserving environment strip off a trailing newline
    // since the closing of the opening block element will automatically result
    // in a new block for elements following the <pre>
    // ...foo\n</pre><p>blah -> foo</pre><p>blah
    if ((at(p).wsm == QTextHtmlParserNode::WhiteSpacePre
         || at(p).wsm == QTextHtmlParserNode::WhiteSpacePreWrap)
        && at(p).isBlock()) {
        if (at(last()).text.endsWith(QLatin1Char('\n')))
            nodes[last()].text.chop(1);
    }

    newNode(at(p).parent);
    resolveNode();
}

// parses a tag beginning with "!"
void QTextHtmlParser::parseExclamationTag()
{
    ++pos;
    if (hasPrefix(QLatin1Char('-'),1) && hasPrefix(QLatin1Char('-'),2)) {
        pos += 3;
        // eat comments
        int end = txt.indexOf(QLatin1String("-->"), pos);
        pos = (end >= 0 ? end + 3 : len);
    } else {
        // eat internal tags
        while (pos < len) {
            QChar c = txt.at(pos++);
            if (c == QLatin1Char('>'))
                break;
        }
    }
}

// parses an entity after "&", and returns it
QString QTextHtmlParser::parseEntity()
{
    int recover = pos;
    QString entity;
    while (pos < len) {
        QChar c = txt.at(pos++);
        if (c.isSpace() || pos - recover > 8) {
            goto error;
        }
        if (c == QLatin1Char(';'))
            break;
        entity += c;
    }
    {
        QChar resolved = resolveEntity(entity);
        if (!resolved.isNull())
            return QString(resolved);
    }
    if (entity.length() > 1 && entity.at(0) == QLatin1Char('#')) {
        entity.remove(0, 1); // removing leading #

        int base = 10;
        bool ok = false;

        if (entity.at(0).toLower() == QLatin1Char('x')) { // hex entity?
            entity.remove(0, 1);
            base = 16;
        }

        uint uc = entity.toUInt(&ok, base);
        if (ok) {
            if (uc >= 0x80 && uc < 0x80 + (sizeof(windowsLatin1ExtendedCharacters)/sizeof(windowsLatin1ExtendedCharacters[0])))
                uc = windowsLatin1ExtendedCharacters[uc - 0x80];
            QString str;
            if (uc > 0xffff) {
                // surrogate pair
                uc -= 0x10000;
                ushort high = uc/0x400 + 0xd800;
                ushort low = uc%0x400 + 0xdc00;
                str.append(QChar(high));
                str.append(QChar(low));
            } else {
                str.append(QChar(uc));
            }
            return str;
        }
    }
error:
    pos = recover;
    return QLatin1String("&");
}

// parses one word, possibly quoted, and returns it
QString QTextHtmlParser::parseWord()
{
    QString word;
    if (hasPrefix(QLatin1Char('\"'))) { // double quotes
        ++pos;
        while (pos < len) {
            QChar c = txt.at(pos++);
            if (c == QLatin1Char('\"'))
                break;
            else if (c == QLatin1Char('&'))
                word += parseEntity();
            else
                word += c;
        }
    } else if (hasPrefix(QLatin1Char('\''))) { // single quotes
        ++pos;
        while (pos < len) {
            QChar c = txt.at(pos++);
            if (c == QLatin1Char('\''))
                break;
            else
                word += c;
        }
    } else { // normal text
        while (pos < len) {
            QChar c = txt.at(pos++);
            if (c == QLatin1Char('>')
                || (c == QLatin1Char('/') && hasPrefix(QLatin1Char('>'), 1))
                || c == QLatin1Char('<')
                || c == QLatin1Char('=')
                || c.isSpace()) {
                --pos;
                break;
            }
            if (c == QLatin1Char('&'))
                word += parseEntity();
            else
                word += c;
        }
    }
    return word;
}

// gives the new node the right parent
QTextHtmlParserNode *QTextHtmlParser::resolveParent()
{
    QTextHtmlParserNode *node = &nodes.last();

    int p = node->parent;

    // Excel gives us buggy HTML with just tr without surrounding table tags
    // or with just td tags

    if (node->id == Html_td) {
        int n = p;
        while (n && at(n).id != Html_tr)
            n = at(n).parent;

        if (!n) {
            nodes.insert(nodes.count() - 1, QTextHtmlParserNode());
            nodes.insert(nodes.count() - 1, QTextHtmlParserNode());

            QTextHtmlParserNode *table = &nodes[nodes.count() - 3];
            table->parent = p;
            table->id = Html_table;
            table->tag = QLatin1String("table");
            table->children.append(nodes.count() - 2); // add row as child

            QTextHtmlParserNode *row = &nodes[nodes.count() - 2];
            row->parent = nodes.count() - 3; // table as parent
            row->id = Html_tr;
            row->tag = QLatin1String("tr");

            p = nodes.count() - 2;
            node = &nodes.last(); // re-initialize pointer
        }
    }

    if (node->id == Html_tr) {
        int n = p;
        while (n && at(n).id != Html_table)
            n = at(n).parent;

        if (!n) {
            nodes.insert(nodes.count() - 1, QTextHtmlParserNode());
            QTextHtmlParserNode *table = &nodes[nodes.count() - 2];
            table->parent = p;
            table->id = Html_table;
            table->tag = QLatin1String("table");
            p = nodes.count() - 2;
            node = &nodes.last(); // re-initialize pointer
        }
    }

    // permit invalid html by letting block elements be children
    // of inline elements with the exception of paragraphs:
    //
    // a new paragraph closes parent inline elements (while loop),
    // unless they themselves are children of a non-paragraph block
    // element (if statement)
    //
    // For example:
    //
    // <body><p><b>Foo<p>Bar <-- second <p> implicitly closes <b> that
    //                           belongs to the first <p>. The self-nesting
    //                           check further down prevents the second <p>
    //                           from nesting into the first one then.
    //                           so Bar is not bold.
    //
    // <body><b><p>Foo <-- Foo should be bold.
    //
    // <body><b><p>Foo<p>Bar <-- Foo and Bar should be bold.
    //
    if (node->id == Html_p) {
        while (p && !at(p).isBlock())
            p = at(p).parent;

        if (!p || at(p).id != Html_p)
            p = node->parent;
    }

    // some elements are not self nesting
    if (node->id == at(p).id
        && node->isNotSelfNesting())
        p = at(p).parent;

    // some elements are not allowed in certain contexts
    while (p && !node->allowedInContext(at(p).id)
           // ### make new styles aware of empty tags
           || at(p).mayNotHaveChildren()
       ) {
        p = at(p).parent;
    }

    node->parent = p;

    // makes it easier to traverse the tree, later
    nodes[p].children.append(nodes.count() - 1);
    return node;
}

// sets all properties on the new node
void QTextHtmlParser::resolveNode()
{
    QTextHtmlParserNode *node = &nodes.last();
    const QTextHtmlParserNode *parent = &nodes.at(node->parent);
    node->initializeProperties(parent, this);
}

bool QTextHtmlParserNode::isNestedList(const QTextHtmlParser *parser) const
{
    if (!isListStart())
        return false;

    int p = parent;
    while (p) {
        if (parser->at(p).isListStart())
            return true;
        p = parser->at(p).parent;
    }
    return false;
}

void QTextHtmlParserNode::initializeProperties(const QTextHtmlParserNode *parent, const QTextHtmlParser *parser)
{
    // inherit properties from parent element
    charFormat = parent->charFormat;

    if (parent->displayMode == QTextHtmlElement::DisplayNone)
        displayMode = QTextHtmlElement::DisplayNone;

    if (parent->id != Html_table || id == Html_caption) {
        if (parent->blockFormat.hasProperty(QTextFormat::BlockAlignment))
            blockFormat.setAlignment(parent->blockFormat.alignment());
        else
            blockFormat.clearProperty(QTextFormat::BlockAlignment);
    }
    // we don't paint per-row background colors, yet. so as an
    // exception inherit the background color here
    // we also inherit the background between inline elements
    if ((parent->id != Html_tr || !isTableCell())
        && (displayMode != QTextHtmlElement::DisplayInline || parent->displayMode != QTextHtmlElement::DisplayInline)) {
        charFormat.clearProperty(QTextFormat::BackgroundBrush);
    }

    listStyle = parent->listStyle;
    // makes no sense to inherit that property, a named anchor is a single point
    // in the document, which is set by the DocumentFragment
    charFormat.clearProperty(QTextFormat::AnchorName);
    wsm = parent->wsm;

    // initialize remaining properties
    margin[QTextHtmlParser::MarginLeft] = 0;
    margin[QTextHtmlParser::MarginRight] = 0;
    margin[QTextHtmlParser::MarginTop] = 0;
    margin[QTextHtmlParser::MarginBottom] = 0;
    cssFloat = QTextFrameFormat::InFlow;

    // set element specific attributes
    switch (id) {
        case Html_a:
            charFormat.setAnchor(true);
            for (int i = 0; i < attributes.count(); i += 2) {
                const QString key = attributes.at(i);
                if (key.compare(QLatin1String("href"), Qt::CaseInsensitive) == 0
                    && !attributes.at(i + 1).isEmpty()) {
                    hasHref = true;
                    charFormat.setUnderlineStyle(QTextCharFormat::SingleUnderline);
                    charFormat.setForeground(Qt::blue);
                }
            }

            break;
        case Html_em:
        case Html_i:
        case Html_cite:
        case Html_address:
        case Html_var:
        case Html_dfn:
            charFormat.setFontItalic(true);
            break;
        case Html_big:
            charFormat.setProperty(QTextFormat::FontSizeAdjustment, int(1));
            break;
        case Html_small:
            charFormat.setProperty(QTextFormat::FontSizeAdjustment, int(-1));
            break;
        case Html_strong:
        case Html_b:
            charFormat.setFontWeight(QFont::Bold);
            break;
        case Html_h1:
            charFormat.setFontWeight(QFont::Bold);
            charFormat.setProperty(QTextFormat::FontSizeAdjustment, int(3));
            margin[QTextHtmlParser::MarginTop] = 18;
            margin[QTextHtmlParser::MarginBottom] = 12;
            break;
        case Html_h2:
            charFormat.setFontWeight(QFont::Bold);
            charFormat.setProperty(QTextFormat::FontSizeAdjustment, int(2));
            margin[QTextHtmlParser::MarginTop] = 16;
            margin[QTextHtmlParser::MarginBottom] = 12;
            break;
        case Html_h3:
            charFormat.setFontWeight(QFont::Bold);
            charFormat.setProperty(QTextFormat::FontSizeAdjustment, int(1));
            margin[QTextHtmlParser::MarginTop] = 14;
            margin[QTextHtmlParser::MarginBottom] = 12;
            break;
        case Html_h4:
            charFormat.setFontWeight(QFont::Bold);
            charFormat.setProperty(QTextFormat::FontSizeAdjustment, int(0));
            margin[QTextHtmlParser::MarginTop] = 12;
            margin[QTextHtmlParser::MarginBottom] = 12;
            break;