C++ QWebElement::toOuterXml方法代码示例

void NoteFormatter::modifyPdfTags(qint32 resLid, QWebElement &enmedia) {
    enmedia.setAttribute("width", "100%");
    enmedia.setAttribute("height", "100%");
    enmedia.setAttribute("lid", QString::number(resLid));
    QString x = enmedia.toOuterXml();
    x.replace("en-media", "object");
    enmedia.setOuterXml(x);
    x = enmedia.toOuterXml();
}

void MainWindow::saveSelected(const QString &filename)
{
	if (_analyses->count() > 0)
	{
		QWebElement element =  ui->webViewResults->page()->mainFrame()->documentElement();
		QString qHTML = element.toOuterXml();
		_package->analysesHTML = fq(qHTML);

		Json::Value analysesData = Json::arrayValue;
		for (Analyses::iterator itr = _analyses->begin(); itr != _analyses->end(); itr++)
		{
			Analysis *analysis = *itr;
			if (analysis != NULL && analysis->visible())
			{
				Json::Value analysisData = analysis->asJSON();
				analysisData["options"] = analysis->options()->asJSON();
				analysesData.append(analysisData);
			}
		}

		_package->analysesData = analysesData;
		_package->hasAnalyses = true;
	}

	_loader.save(filename, _package);
	_alert->show();
}

void HtmlExtractor::move(QWebElement &element, const QString &file, bool outer)
{
    if(!element.isNull())
    {
        WriteFile(file, outer?element.toOuterXml():element.toInnerXml());
        element.removeFromDocument();
    }
}

void EnmlFormatter::processTodo(QWebElement &node) {
    bool checked=false;
    if (node.hasAttribute("checked"))
        checked = true;
    node.removeAttribute("style");
    node.removeAttribute("type");
    removeInvalidAttributes(node);
    if (checked)
        node.setAttribute("checked", "true");
    node.setOuterXml(node.toOuterXml().replace("<input", "<en-todo").replace("</input", "</en-todo"));
}

// Modify the en-to tag into an input field
void NoteFormatter::modifyTodoTags(QWebElement &todo) {
    todo.setAttribute("type", "checkbox");

    // Checks the en-to tag wheter or not the todo-item is checked or not
    // and sets up the HTML to keep storing the information in value
    QString checked = todo.attribute("checked");
    if (checked.toLower() == "true")
        todo.setAttribute("checked", "checked");
    else
        todo.removeAttribute("checked");

    todo.setAttribute("onClick", "if(!checked) removeAttribute('checked'); else setAttribute('checked', 'checked'); editorWindow.editAlert();");
    todo.setAttribute("style", "cursor: hand;");
    todo.setOuterXml(todo.toOuterXml().replace("en-todo","input"));
}

void EnmlFormatter::fixImgNode(QWebElement &e) {
    QString enType = e.attribute("en-tag", "");

    // Check if we have an en-crypt tag.  Change it from an img to en-crypt
    if (enType.toLower() == "en-crypt") {
        QString encrypted = e.attribute("alt");
        QString cipher = e.attribute("cipher", "RC2");
        QString hint = e.attribute("hint", "");
        QString length = e.attribute("length", "64");
        e.removeAttribute("onmouseover");
        e.removeAttribute("name");
        e.removeAttribute("alt");
        e.removeAttribute("en-tag");
        e.removeAttribute("contenteditable");
        e.removeAttribute("style");
        removeInvalidAttributes(e);
        e.setInnerXml(encrypted);
        e.setOuterXml("<en-crypt cipher=\"" +cipher +"\" length=\"" +
                      length + "\" hint=\"" +hint
                      +"\">"+encrypted+"</en-crypt>");
//        e.setOuterXml(e.toOuterXml().replace("<img", "<en-crypt").replace("</img", "</en-crypt"));
        return;
    }

    // Check if we have a temporary image.  If so, remove it
    if (enType.toLower() ==  "temporary") {;
        e.removeFromDocument();
        return;
    }


    // Latex images are really just img tags, so we now handle them later
    // Check if we have a LaTeX image.  Remove the parent link tag
//    if (enType.toLower() ==  "en-latex") {
//        enType = "en-media";
//        parent.parentNode().replaceChild(e, parent);
//    }


    // If we've gotten this far, we have an en-media tag
    e.removeAttribute("en-tag");
    int lid = e.attribute("lid").toInt();
    resources.append(lid);
    removeInvalidAttributes(e);
    e.setOuterXml(e.toOuterXml().replace("<img", "<en-media").replace("</img", "</en-media"));
}

void EnmlFormatter::fixObjectNode(QWebElement &e) {
    QString type = e.attribute("type", "");
    if (type == "application/pdf") {
        qint32 lid = e.attribute("lid", "0").toInt();
        e.removeAttribute("width");
        e.removeAttribute("height");
        e.removeAttribute("lid");
        e.removeAttribute("border");
        if (lid>0) {
            resources.append(lid);
             e.setOuterXml(e.toOuterXml().replace("<object", "<en-media").replace("</object", "</en-media"));
        }
        removeInvalidAttributes(e);
    } else {
        e.removeFromDocument();
    }
}

/* If we have an ink note, then we need to pull the image and display it */
bool NoteFormatter::buildInkNote(QWebElement &docElem, QString &hash) {

    ResourceTable resTable(global.db);
    qint32 resLid = resTable.getLidByHashHex(note.guid, hash);
    if (resLid <= 0)
        return false;
    docElem.setAttribute("en-tag", "en-media");
    docElem.setAttribute("lid", QString::number(resLid));
    docElem.setAttribute("type", "application/vnd.evernote.ink");
    QString filename = QString("file:///") +global.fileManager.getDbaDirPath()+QString::number(resLid)+QString(".png");
    docElem.setAttribute("src", filename);
    QString k  = docElem.toOuterXml();
    k.replace("<en-media", "<img");
    k.replace("enmedia>", "img>");;
    docElem.setOuterXml(k);

    return true;
}

QStringList EnmlFormatter::findAllTags(QWebElement &element) {
    QStringList tags;
    QString body = element.toOuterXml();
    body = body.replace("</","<").replace("/>"," ").replace(">", " ");
    int i = body.indexOf("<body")+1
            ;
    // Look throug and get a list of all possible tags
    i = body.indexOf("<", i);

    while (i>0) {
        int eot = body.indexOf(" ",i);
        QString tag = body.mid(i+1, eot-i-1);
        if (!tags.contains(tag) && tag != "body")
            tags.append(tag);
        i=body.indexOf("<", eot);
        if (tag == "body")
            i=-1;
    }
    return tags;
}

void EnmlFormatter::fixLinkNode(QWebElement e) {
    QString enTag = e.attribute("en-tag", "");
    if (enTag.toLower() == "en-media") {
        resources.append(e.attribute("lid").toInt());
        e.removeAttribute("style");
        e.removeAttribute("href");
        e.removeAttribute("title");
        removeInvalidAttributes(e);
        e.removeAllChildren();
        QString newXml = e.toOuterXml();
        newXml.replace("<a", "<en-media");
        newXml.replace("</a>", "</en-media>");
        e.setOuterXml(newXml);
    }
    QString latex = e.attribute("href", "");
    if (latex.toLower().startsWith("latex://")) {
        removeInvalidAttributes(e);
        e.removeAttribute("title");
        e.removeAttribute("href");
        e.setOuterXml(e.toInnerXml());
    }
    removeInvalidAttributes(e);
}

/* Take the ENML note and transform it into HTML that WebKit will
  not complain about */
QByteArray EnmlFormatter::rebuildNoteEnml() {
    resources.clear();
    QByteArray b;
    qint32 index;

    content.replace("</input>","");

    // Strip off HTML header
    index = content.indexOf("<body");
    index = content.indexOf(">", index)+1;
    content.remove(0,index);
    index = content.indexOf("</body");
    content.truncate(index);
    b.append("<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n");
    b.append("<!DOCTYPE en-note SYSTEM 'http://xml.evernote.com/pub/enml2.dtd'>\n");
    b.append("<html><head><title></title></head>");
    b.append("<body style=\"word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space;\" >");
    b.append(content);
    b.append("</body></html>");
    content.clear();
    content = b;

    // Run it through "tidy".  It is a program which will fix any invalid HTML
    // and give us the results back through stdout.  In a perfect world this
    // wouldn't be needed, but WebKit doesn't always give back good HTML.
    QProcess tidyProcess;
    tidyProcess.start("tidy -raw -asxhtml -q -m -u -utf8 ", QIODevice::ReadWrite|QIODevice::Unbuffered);
    QLOG_DEBUG() << "Starting tidy " << tidyProcess.waitForStarted();
    tidyProcess.waitForStarted();
    tidyProcess.write(content);
    tidyProcess.closeWriteChannel();
    tidyProcess.waitForFinished();
    QLOG_DEBUG() << "Stopping tidy " << tidyProcess.waitForFinished() << " Return Code: " << tidyProcess.state();
    QLOG_DEBUG() << "Tidy Errors:" << tidyProcess.readAllStandardError();
    content.clear();
    content.append(tidyProcess.readAllStandardOutput());
    if (content == "") {
        formattingError = true;
        return "";
    }

    // Tidy puts this in place, but we don't really need it.
    content.replace("<form>", "");
    content.replace("</form>", "");

    index = content.indexOf("<body");
    content.remove(0,index);
    content.prepend("<style>img { height:auto; width:auto; max-height:auto; max-width:100%; }</style>");
    content.prepend("<head><meta http-equiv=\"content-type\" content=\"text-html; charset=utf-8\"></head>");
    content.prepend("<html>");
    content.append("</html>");
    content = fixEncryptionTags(content);

    QWebPage page;
    QEventLoop loop;
    page.mainFrame()->setContent(content);
    QObject::connect(&page, SIGNAL(loadFinished(bool)), &loop, SLOT(quit()));


    QWebElement element = page.mainFrame()->documentElement();
    QStringList tags = findAllTags(element);

    for (int i=0; i<tags.size(); i++) {
        QString tag = tags[i];
        QWebElementCollection anchors = page.mainFrame()->findAllElements(tag);
        foreach (QWebElement element, anchors) {
            if (element.tagName().toLower() == "input") {
                processTodo(element);
            } else if (element.tagName().toLower() == "a") {
                fixLinkNode(element);
            } else if (element.tagName().toLower() == "object") {
                fixObjectNode(element);
            } else if (element.tagName().toLower() == "img") {
                fixImgNode(element);
            } else if (!isElementValid(element.tagName()))
                element.removeFromDocument();
        }
    }
    content.clear();
    content.append(element.toOuterXml());

    // Strip off HTML header
    index = content.indexOf("<body");
    index = content.indexOf(">", index)+1;
    content.remove(0,index);
    index = content.indexOf("</body");
    content.truncate(index);
    b.clear();
    b.append("<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n");
    b.append("<!DOCTYPE en-note SYSTEM 'http://xml.evernote.com/pub/enml2.dtd'>");
    b.append("<en-note style=\"word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space;\" >");
    b.append(content);
    b.append("</en-note>");
    content.clear();
    content = b;

    postXmlFix();
    return content;
//.........这里部分代码省略.........

// Modify the en-media tag into an attachment
void NoteFormatter::modifyApplicationTags(QWebElement &enmedia, QString &hash, QString appl) {
    if (appl.toLower() == "vnd.evernote.ink") {
            inkNote = true;
            readOnly = true;
            buildInkNote(enmedia, hash);
            return;
    }
    ResourceTable resTable(global.db);
    QString contextFileName;
    qint32 resLid = resTable.getLidByHashHex(note.guid, hash);
    Resource r;
    resTable.get(r, resLid, false);
    if (!r.data.isSet())
        resourceError = true;
    else {

        // If we are running the formatter and we are not generating a thumbnail
        QString mimetype = "";
        if (r.mime.isSet())
            mimetype = r.mime;
        if (mimetype == "application/pdf" && pdfPreview && !thumbnail) {
           modifyPdfTags(resLid, enmedia);
           return;
        }

        // If we are running the formatter so we can generate a thumbnail and it is a PDF
        if (mimetype == "application/pdf" && pdfPreview && thumbnail) {
            QString printImageFile = global.fileManager.getTmpDirPath() + QString::number(resLid) +QString("-print.jpg");
            QString file = global.fileManager.getDbaDirPath() + QString::number(resLid) +".pdf";
            Poppler::Document *doc;
            doc = Poppler::Document::load(file);
            if (doc == NULL)
                return;

            QImage *image = new QImage(doc->page(0)->renderToImage());
            image->save(printImageFile,"jpg");
            delete image;

            enmedia.setAttribute("src", printImageFile);
            enmedia.removeAttribute("hash");
            enmedia.removeAttribute("type");
            enmedia.setOuterXml(enmedia.toOuterXml().replace("<en-media","<img"));
            enmedia.setOuterXml(enmedia.toOuterXml().replace("</en-media>","</img>"));
            return;
        }
        QString fileDetails = "";
        MimeReference ref;
        ResourceAttributes attributes;
        if (r.attributes.isSet())
            attributes = r.attributes;
        if (attributes.fileName.isSet())
            fileDetails = ref.getExtensionFromMime(r.mime, fileDetails);

        enmedia.setAttribute("href", QString("nnres:") +global.fileManager.getDbaDirPath()+QString::number(resLid)
                             +fileDetails);
        contextFileName = global.fileManager.getTmpDirPath("")+QString::number(resLid) +global.attachmentNameDelimeter + fileDetails;

        // Setup the context menu.  This is useful if we want to do a "save as" or such
        contextFileName = contextFileName.replace("\\", "/");
        enmedia.setAttribute("onContextMenu", "window.browserWindow.resourceContextMenu('" +contextFileName +"');");
        enmedia.setAttribute("en-tag", "en-media");
        enmedia.setAttribute("lid", QString::number(resLid));

        enmedia.appendInside("<img/>");
        QWebElement newText = enmedia.lastChild();

        // Build an icon of the image
        QString fileExt;
        if (attributes.fileName.isSet())
            fileExt = attributes.fileName;
        else
            fileExt = appl;
        QString fn;
        QString mime;
        if (attributes.fileName.isSet())
            fn = attributes.fileName;
        if (r.mime.isSet())
            mime = r.mime;
        fileExt = ref.getExtensionFromMime(mime, fn);
        QString icon = findIcon(resLid, r, fileExt);
        newText.setAttribute("src", "file:///"+icon);
        if (attributes.fileName.isSet())
            newText.setAttribute("title",attributes.fileName);
        newText.setAttribute("en-tag", "temporary");
        //Rename the tag to a <a> link
        enmedia.setOuterXml(enmedia.toOuterXml().replace("<en-media","<a"));
        enmedia.setOuterXml(enmedia.toOuterXml().replace("</en-media>","</a>"));
    }
}

/* Modify an image tag.  Basically we turn it back into a picture, write out the file, and
  modify the ENML */
void NoteFormatter::modifyImageTags(QWebElement &enMedia, QString &hash) {
    QString mimetype = enMedia.attribute("type");
    qint32 resLid = 0;
    resLid = hashMap[hash];
    QString highlightString = "";
    if (resLid>0) {
        QLOG_TRACE() << "Getting resource";
        Resource r = resourceMap[resLid];
        QLOG_TRACE() << "resource retrieved";
        MimeReference ref;
        QString filename;
        ResourceAttributes attributes;
        if (r.attributes.isSet())
            attributes = r.attributes;
        if (attributes.fileName.isSet())
            filename = attributes.fileName;
        QString type = ref.getExtensionFromMime(mimetype, filename);

        Data data;
        if (r.data.isSet())
            data = r.data;
        if (data.size.isSet() && data.size > 0) {
            QString imgfile = "file:///"+global.fileManager.getDbDirPath(QString("dba/") +QString::number(resLid) +type);
            enMedia.setAttribute("src", imgfile);
            // Check if this is a LaTeX image
            ResourceAttributes attributes;
            if (r.attributes.isSet())
                attributes = r.attributes;
            QString sourceUrl = "";
            if (attributes.sourceURL.isSet())
                sourceUrl = attributes.sourceURL;
            if (sourceUrl.toLower().startsWith("http://latex.codecogs.com/gif.latex?")) {
                enMedia.appendInside("<img/>");
                QWebElement newText = enMedia.lastChild();
                enMedia.setAttribute("en-tag", "en-latex");
                newText.setAttribute("onMouseOver", "style.cursor='pointer'");
                newText.setAttribute("title", sourceUrl);
                newText.setAttribute("href", "latex://"+QString::number(resLid));
            }
            enMedia.setAttribute("onContextMenu", "window.browserWindow.imageContextMenu('"
                                 +QString::number(resLid) +"', '"
                                 +QString::number(resLid) +type  +"');");
            highlightString = addImageHighlight(resLid, imgfile);
            if (highlightString != "")
                enMedia.setAttribute("onload", highlightString);

        }
    } else {
        resourceError = true;
        readOnly = true;
    }

    // Reset the tags to something that WebKit will understand
    enMedia.setAttribute("en-tag", "en-media");
    enMedia.setPlainText("");
    enMedia.setAttribute("lid", QString::number(resLid));

    // rename the <enmedia> tag to <img>
    enMedia.setOuterXml(enMedia.toOuterXml().replace("<en-media","<img"));
    enMedia.setOuterXml(enMedia.toOuterXml().replace("</en-media>","</img>"));
}

/*
  This will go through and modify some of the ENML tags and turn
  them into HTML tags.  Things like en-media & en-crypt have no
  HTML values, so we turn them into HTML.
  */
void NoteFormatter::modifyTags(QWebPage &doc) {
    tempFiles.clear();

    // Modify en-media tags
    QLOG_TRACE() << "Searching for all en-media tags;";
    QWebElementCollection anchors = doc.mainFrame()->findAllElements("en-media");
    QLOG_TRACE() << "Search complete: " << anchors.toList().size();
    foreach (QWebElement enmedia, anchors) {
        if (enmedia.hasAttribute("type")) {
            QString attr = enmedia.attribute("type");
            QString hash = enmedia.attribute("hash");
            QStringList type = attr.split("/");
            if (type.size() >= 2) {
                QString appl = type[1];
                QLOG_TRACE() << "En-Media tag type: " << type[0];
                if (type[0] == "image")
                    modifyImageTags(enmedia, hash);
                else
                    modifyApplicationTags(enmedia, hash, appl);
                QLOG_TRACE() << "Type modified";
            }
        }
    }

    // Modify todo tags
    anchors = doc.mainFrame()->findAllElements("en-todo");
    qint32 enTodoCount = anchors.count();
    for (qint32 i=enTodoCount-1; i>=0; i--) {
            QWebElement enmedia = anchors.at(i);
            modifyTodoTags(enmedia);
    }

    anchors = doc.mainFrame()->findAllElements("en-crypt");
    qint32 enCryptLen = anchors.count();
    for (qint32 i=enCryptLen-1; i>=0; i--) {
        QWebElement enmedia = anchors.at(i);
        QString hint = enmedia.attribute("hint");
        QString cipher = enmedia.attribute("cipher", "RC2");
        QString length = enmedia.attribute("length","64");

        enmedia.setAttribute("contentEditable","false");
        enmedia.setAttribute("src", QString("file://")+global.fileManager.getImageDirPath("encrypt.png"));
        enmedia.setAttribute("en-tag","en-crypt");
        enmedia.setAttribute("cipher", cipher);
        enmedia.setAttribute("length", length);
        enmedia.setAttribute("hint", hint);
        enmedia.setAttribute("alt", enmedia.toInnerXml());
        global.cryptCounter++;
        enmedia.setAttribute("id", "crypt"+QString().number(global.cryptCounter));
        QString encryptedText = enmedia.toInnerXml();

        // If the encryption string contains crlf at the end, remove them because they mess up the javascript.
        if (encryptedText.endsWith("\n"))
                encryptedText.truncate(encryptedText.length()-1);
        if (encryptedText.endsWith("\r"))
                encryptedText.truncate(encryptedText.length()-1);

        // Add the commands
        hint = hint.replace("'","&apos;");
        enmedia.setAttribute("onClick", "window.browserWindow.decryptText('crypt"+
                             QString().number(global.cryptCounter)+
                             "', '"+encryptedText+"', '"+
                             hint +"', '" +
                             cipher+ "', " +
                             length +
                             ");");
        enmedia.setAttribute("onMouseOver", "style.cursor='hand'");
        enmedia.setInnerXml("");
        QString k = enmedia.toOuterXml();
        k.replace("<en-crypt", "<img");
        k.replace("img>", "<en-crypt");
        enmedia.setOuterXml(k);
    }


    // Modify link tags
    anchors = doc.mainFrame()->findAllElements("a");
    enCryptLen = anchors.count();
    for (qint32 i=0; i<anchors.count(); i++) {
        QWebElement element = anchors.at(i);
        if (!element.attribute("href").toLower().startsWith("latex://"))
            element.setAttribute("title", element.attribute("href"));
        else {
            element.setAttribute("title", element.attribute("title").toLower().replace("http://latex.codecogs.com/gif.latex?",""));
        }
    }

}