qtexthtmlparser.cpp

奇趣公司比较新的qt/emd版本

    0x0161, // 0x9A
    0x203A, // 0x9B
    0x0153, // 0x9C
    0x009D, // 0x9D direct mapping
    0x017E, // 0x9E
    0x0178  // 0x9F
};

// the displayMode value is according to the what are blocks in the piecetable, not
// what the w3c defines.
static const QTextHtmlElement elements[Html_NumElements]= {
    { "a",          Html_a,          QTextHtmlElement::DisplayInline },
    { "address",    Html_address,    QTextHtmlElement::DisplayInline },
    { "b",          Html_b,          QTextHtmlElement::DisplayInline },
    { "big",        Html_big,        QTextHtmlElement::DisplayInline },
    { "blockquote", Html_blockquote, QTextHtmlElement::DisplayBlock },
    { "body",       Html_body,       QTextHtmlElement::DisplayBlock },
    { "br",         Html_br,         QTextHtmlElement::DisplayInline },
    { "caption",    Html_caption,    QTextHtmlElement::DisplayBlock },
    { "center",     Html_center,     QTextHtmlElement::DisplayBlock },
    { "cite",       Html_cite,       QTextHtmlElement::DisplayInline },
    { "code",       Html_code,       QTextHtmlElement::DisplayInline },
    { "dd",         Html_dd,         QTextHtmlElement::DisplayBlock },
    { "dfn",        Html_dfn,        QTextHtmlElement::DisplayInline },
    { "div",        Html_div,        QTextHtmlElement::DisplayBlock },
    { "dl",         Html_dl,         QTextHtmlElement::DisplayBlock },
    { "dt",         Html_dt,         QTextHtmlElement::DisplayBlock },
    { "em",         Html_em,         QTextHtmlElement::DisplayInline },
    { "font",       Html_font,       QTextHtmlElement::DisplayInline },
    { "h1",         Html_h1,         QTextHtmlElement::DisplayBlock },
    { "h2",         Html_h2,         QTextHtmlElement::DisplayBlock },
    { "h3",         Html_h3,         QTextHtmlElement::DisplayBlock },
    { "h4",         Html_h4,         QTextHtmlElement::DisplayBlock },
    { "h5",         Html_h5,         QTextHtmlElement::DisplayBlock },
    { "h6",         Html_h6,         QTextHtmlElement::DisplayBlock },
    { "head",       Html_head,       QTextHtmlElement::DisplayNone },
    { "hr",         Html_hr,         QTextHtmlElement::DisplayBlock },
    { "html",       Html_html,       QTextHtmlElement::DisplayInline },
    { "i",          Html_i,          QTextHtmlElement::DisplayInline },
    { "img",        Html_img,        QTextHtmlElement::DisplayInline },
    { "kbd",        Html_kbd,        QTextHtmlElement::DisplayInline },
    { "li",         Html_li,         QTextHtmlElement::DisplayBlock },
    { "link",       Html_link,       QTextHtmlElement::DisplayNone },
    { "meta",       Html_meta,       QTextHtmlElement::DisplayNone },
    { "nobr",       Html_nobr,       QTextHtmlElement::DisplayInline },
    { "ol",         Html_ol,         QTextHtmlElement::DisplayBlock },
    { "p",          Html_p,          QTextHtmlElement::DisplayBlock },
    { "pre",        Html_pre,        QTextHtmlElement::DisplayBlock },
    { "qt",         Html_body /*deliberate mapping*/, QTextHtmlElement::DisplayBlock },
    { "s",          Html_s,          QTextHtmlElement::DisplayInline },
    { "samp",       Html_samp,       QTextHtmlElement::DisplayInline },
    { "small",      Html_small,      QTextHtmlElement::DisplayInline },
    { "span",       Html_span,       QTextHtmlElement::DisplayInline },
    { "strong",     Html_strong,     QTextHtmlElement::DisplayInline },
    { "style",      Html_style,      QTextHtmlElement::DisplayNone },
    { "sub",        Html_sub,        QTextHtmlElement::DisplayInline },
    { "sup",        Html_sup,        QTextHtmlElement::DisplayInline },
    { "table",      Html_table,      QTextHtmlElement::DisplayTable },
    { "tbody",      Html_tbody,      QTextHtmlElement::DisplayTable },
    { "td",         Html_td,         QTextHtmlElement::DisplayBlock },
    { "tfoot",      Html_tfoot,      QTextHtmlElement::DisplayTable },
    { "th",         Html_th,         QTextHtmlElement::DisplayBlock },
    { "thead",      Html_thead,      QTextHtmlElement::DisplayTable },
    { "title",      Html_title,      QTextHtmlElement::DisplayNone },
    { "tr",         Html_tr,         QTextHtmlElement::DisplayTable },
    { "tt",         Html_tt,         QTextHtmlElement::DisplayInline },
    { "u",          Html_u,          QTextHtmlElement::DisplayInline },
    { "ul",         Html_ul,         QTextHtmlElement::DisplayBlock },
    { "var",        Html_var,        QTextHtmlElement::DisplayInline },
};


static bool operator<(const QString &str, const QTextHtmlElement &e)
{
    return str < QLatin1String(e.name);
}

static bool operator<(const QTextHtmlElement &e, const QString &str)
{
    return QLatin1String(e.name) < str;
}

static const QTextHtmlElement *lookupElement(const QString &element)
{
    const QTextHtmlElement *start = &elements[0];
    const QTextHtmlElement *end = &elements[Html_NumElements];
    const QTextHtmlElement *e = qBinaryFind(start, end, element);
    if (e == end)
        return 0;
    return e;
}

int QTextHtmlParser::lookupElement(const QString &element)
{
    const QTextHtmlElement *e = ::lookupElement(element);
    if (!e)
        return -1;
    return e->id;
}

// quotes newlines as "\\n"
static QString quoteNewline(const QString &s)
{
    QString n = s;
    n.replace(QLatin1Char('\n'), QLatin1String("\\n"));
    return n;
}

QTextHtmlParserNode::QTextHtmlParserNode()
    : parent(0), id(Html_unknown),
      cssFloat(QTextFrameFormat::InFlow), hasOwnListStyle(false),
      hasCssListIndent(false), isEmptyParagraph(false), isTextFrame(false), isRootFrame(false),
      displayMode(QTextHtmlElement::DisplayInline), hasHref(false),
      listStyle(QTextListFormat::ListStyleUndefined), imageWidth(-1), imageHeight(-1), tableBorder(0),
      tableCellRowSpan(1), tableCellColSpan(1), tableCellSpacing(2), tableCellPadding(0),
      borderBrush(Qt::darkGray), borderStyle(QTextFrameFormat::BorderStyle_Outset),
      userState(-1), cssListIndent(0), wsm(WhiteSpaceModeUndefined)
{
    margin[QTextHtmlParser::MarginLeft] = 0;
    margin[QTextHtmlParser::MarginRight] = 0;
    margin[QTextHtmlParser::MarginTop] = 0;
    margin[QTextHtmlParser::MarginBottom] = 0;
}

void QTextHtmlParser::dumpHtml()
{
    for (int i = 0; i < count(); ++i) {
        qDebug().nospace() << qPrintable(QString(depth(i)*4, QLatin1Char(' ')))
                           << qPrintable(at(i).tag) << ":"
                           << quoteNewline(at(i).text);
            ;
    }
}

QTextHtmlParserNode *QTextHtmlParser::newNode(int parent)
{
    QTextHtmlParserNode *lastNode = &nodes.last();
    QTextHtmlParserNode *newNode = 0;

    bool reuseLastNode = true;

    if (nodes.count() == 1) {
        reuseLastNode = false;
    } else if (lastNode->tag.isEmpty()) {

        if (lastNode->text.isEmpty()) {
            reuseLastNode = true;
        } else { // last node is a text node (empty tag) with some text

            if (lastNode->text.length() == 1 && lastNode->text.at(0).isSpace()) {

                int lastSibling = count() - 2;
                while (lastSibling
                       && at(lastSibling).parent != lastNode->parent
                       && at(lastSibling).displayMode == QTextHtmlElement::DisplayInline) {
                    lastSibling = at(lastSibling).parent;
                }

                if (at(lastSibling).displayMode == QTextHtmlElement::DisplayInline) {
                    reuseLastNode = false;
                } else {
                    reuseLastNode = true;
                }
            } else {
                // text node with real (non-whitespace) text -> nothing to re-use
                reuseLastNode = false;
            }

        }

    } else {
        // last node had a proper tag -> nothing to re-use
        reuseLastNode = false;
    }

    if (reuseLastNode) {
        newNode = lastNode;
        newNode->tag.clear();
        newNode->text.clear();
        newNode->id = Html_unknown;
    } else {
        nodes.resize(nodes.size() + 1);
        newNode = &nodes.last();
    }

    newNode->parent = parent;
    return newNode;
}

void QTextHtmlParser::parse(const QString &text, const QTextDocument *_resourceProvider)
{
    nodes.clear();
    nodes.resize(1);
    txt = text;
    pos = 0;
    len = txt.length();
    textEditMode = false;
    resourceProvider = _resourceProvider;
    parse();
    //dumpHtml();
}

int QTextHtmlParser::depth(int i) const
{
    int depth = 0;
    while (i) {
        i = at(i).parent;
        ++depth;
    }
    return depth;
}

int QTextHtmlParser::margin(int i, int mar) const {
    int m = 0;
    const QTextHtmlParserNode *node;
    if (mar == MarginLeft
        || mar == MarginRight) {
        while (i) {
            node = &at(i);
            if (!node->isBlock() && node->id != Html_table)
                break;
            if (node->isTableCell())
                break;
            m += node->margin[mar];
            i = node->parent;
        }
    }
    return m;
}

int QTextHtmlParser::topMargin(int i) const
{
    if (!i)
        return 0;
    return at(i).margin[MarginTop];
}

int QTextHtmlParser::bottomMargin(int i) const
{
    if (!i)
        return 0;
    return at(i).margin[MarginBottom];
}

void QTextHtmlParser::eatSpace()
{
    while (pos < len && txt.at(pos).isSpace() && txt.at(pos) != QChar::ParagraphSeparator)
        pos++;
}

void QTextHtmlParser::parse() {
    QTextHtmlParserNode::WhiteSpaceMode wsm = QTextHtmlParserNode::WhiteSpaceNormal;
    while (pos < len) {
        QChar c = txt.at(pos++);
        if (c == QLatin1Char('<')) {
            parseTag();
            wsm = nodes.last().wsm;
        } else if (c == QLatin1Char('&')) {
            nodes.last().text += parseEntity();
        } else {
            nodes.last().text += c;
        }
    }
}

// parses a tag after "<"
void QTextHtmlParser::parseTag()
{
    eatSpace();

    // handle comments and other exclamation mark declarations
    if (hasPrefix(QLatin1Char('!'))) {
        parseExclamationTag();
        if (nodes.last().wsm != QTextHtmlParserNode::WhiteSpacePre
            && nodes.last().wsm != QTextHtmlParserNode::WhiteSpacePreWrap
            && !textEditMode)
            eatSpace();
        return;
    }

    // if close tag just close
    if (hasPrefix(QLatin1Char('/'))) {
        if (nodes.last().id == Html_style) {
            QCss::Parser parser(nodes.last().text);
            QCss::StyleSheet sheet;
            parser.parse(&sheet);
            inlineStyleSheets.append(sheet);
            resolveStyleSheetImports(sheet);
        }
        parseCloseTag();
        return;
    }

    int p = last();
    while (p && at(p).tag.size() == 0)
        p = at(p).parent;

    QTextHtmlParserNode *node = newNode(p);

    // parse tag name
    node->tag = parseWord().toLower();

    const QTextHtmlElement *elem = ::lookupElement(node->tag);
    if (elem) {
        node->id = elem->id;
        node->displayMode = elem->displayMode;
    } else {
        node->id = Html_unknown;
    }

    node->attributes.clear();
    // _need_ at least one space after the tag name, otherwise there can't be attributes
    if (pos < len && txt.at(pos).isSpace())
        node->attributes = parseAttributes();

    // resolveParent() may have to change the order in the tree and
    // insert intermediate nodes for buggy HTML, so re-initialize the 'node'
    // pointer through the return value
    node = resolveParent();
    resolveNode();

    const int nodeIndex = nodes.count() - 1; // this new node is always the last
    node->applyCssDeclarations(declarationsForNode(nodeIndex), resourceProvider);
    applyAttributes(node->attributes);

    // finish tag
    bool tagClosed = false;
    while (pos < len && txt.at(pos) != QLatin1Char('>')) {
        if (txt.at(pos) == QLatin1Char('/'))
            tagClosed = true;

        pos++;
    }
    pos++;

    // in a white-space preserving environment strip off a initial newline
    // since the element itself already generates a newline
    if ((node->wsm == QTextHtmlParserNode::WhiteSpacePre
         || node->wsm == QTextHtmlParserNode::WhiteSpacePreWrap)
        && node->isBlock()) {
        if (pos < len - 1 && txt.at(pos) == QLatin1Char('\n'))
            ++pos;
    }

    if (node->mayNotHaveChildren() || tagClosed) {
        newNode(node->parent);
        resolveNode();
    }
}

// parses a tag beginning with "/"
void QTextHtmlParser::parseCloseTag()
{
    ++pos;
    QString tag = parseWord().toLower().trimmed();
    while (pos < len) {
        QChar c = txt.at(pos++);
        if (c == QLatin1Char('>'))
            break;
    }