Merge pull request #68 from xiaoyifang/fix/iframe

Fix/iframe

article_maker.cc

@@ -733,19 +733,19 @@ void ArticleRequest::bodyFinished()
     {
         update();
         qDebug() << "send dicts(stemmed):" << word << ":" << dictIds;
-        emit GlobalBroadcaster::instance()->emitDictIds(ActiveDictIds{word, dictIds});
+        emit GlobalBroadcaster::instance()->dictionaryChanges(ActiveDictIds{word, dictIds});
         dictIds.clear();
     }
     else {
       finish();
       qDebug() << "send dicts(finished):" << word << ":" << dictIds;
-      emit GlobalBroadcaster::instance()->emitDictIds(ActiveDictIds{word, dictIds});
+      emit GlobalBroadcaster::instance()->dictionaryChanges(ActiveDictIds{word, dictIds});
       dictIds.clear();
     }
   } else if (wasUpdated) {
     update();
     qDebug() << "send dicts(updated):" << word << ":" << dictIds;
-    emit GlobalBroadcaster::instance()->emitDictIds(ActiveDictIds{word, dictIds});
+    emit GlobalBroadcaster::instance()->dictionaryChanges(ActiveDictIds{word, dictIds});
     dictIds.clear();
   }
 }

article_netmgr.cc

@@ -29,7 +29,7 @@ using std::string;
     connect( baseReply, SIGNAL( errorOccurred( QNetworkReply::NetworkError) ),
              this, SLOT( applyError( QNetworkReply::NetworkError ) ) );
 
-    connect( baseReply, SIGNAL( readyRead() ), this, SLOT( readDataFromBase() ) );
+    connect( baseReply, SIGNAL( readyRead() ), this, SIGNAL( readyRead() ) );
 
     // Redirect QNetworkReply signals
 
@@ -68,7 +68,8 @@ using std::string;
     QList< QByteArray > rawHeaders = baseReply->rawHeaderList();
     for( QList< QByteArray >::iterator it = rawHeaders.begin(); it != rawHeaders.end(); ++it )
     {
-      if( it->toLower() != "x-frame-options" )
+      auto headerName = it->toLower();
+      if( headerName != "x-frame-options" && headerName != "content-security-policy")
         setRawHeader( *it, baseReply->rawHeader( *it ) );
     }
 
@@ -121,7 +122,7 @@ using std::string;
 
   qint64 AllowFrameReply::bytesAvailable() const
   {
-    return buffer.size() + QNetworkReply::bytesAvailable();
+    return baseReply->bytesAvailable();
   }
 
   void AllowFrameReply::applyError( QNetworkReply::NetworkError code )
@@ -130,20 +131,22 @@ using std::string;
     emit errorOccurred( code );
   }
 
-  void AllowFrameReply::readDataFromBase()
-  {
-    QByteArray data;
-    data.resize( baseReply->bytesAvailable() );
-    baseReply->read( data.data(), data.size() );
-    buffer += data;
-    emit readyRead();
-  }
+//  void AllowFrameReply::readDataFromBase()
+//  {
+////    QByteArray data;
+////    data.resize( baseReply->bytesAvailable() );
+////    baseReply->read( data.data(), data.size() );
+////    buffer += data;
+//    emit readyRead();
+//  }
 
   qint64 AllowFrameReply::readData( char * data, qint64 maxSize )
   {
-    qint64 size = qMin( maxSize, qint64( buffer.size() ) );
-    memcpy( data, buffer.data(), size );
-    buffer.remove( 0, size );
+    auto bytesAvailable= baseReply->bytesAvailable();
+    qint64 size = qMin( maxSize, bytesAvailable );
+    baseReply->read( data, size );
+//    memcpy( data, buffer.data(), size );
+//    buffer.remove( 0, size );
     return size;
   }
 
@@ -151,6 +154,7 @@ QNetworkReply * ArticleNetworkAccessManager::createRequest( Operation op,
                                                             QNetworkRequest const & req,
                                                             QIODevice * outgoingData )
 {
+  QUrl url;
   if ( op == GetOperation )
   {
     if ( req.url().scheme() == "qrcx" )
@@ -168,7 +172,7 @@ QNetworkReply * ArticleNetworkAccessManager::createRequest( Operation op,
       return QNetworkAccessManager::createRequest( op, newReq, outgoingData );
     }
 
-    QUrl url=req.url();
+    url=req.url();
     QMimeType mineType=db.mimeTypeForUrl (url);
     QString contentType=mineType.name();
 
@@ -178,7 +182,6 @@ QNetworkReply * ArticleNetworkAccessManager::createRequest( Operation op,
             Utils::Url::addQueryItem(url,"word",path.mid(1));
             url.setPath("");
             Utils::Url::addQueryItem(url,"group","1");
-
         }
     }
 
@@ -193,7 +196,7 @@ QNetworkReply * ArticleNetworkAccessManager::createRequest( Operation op,
   //can not match dictionary in the above code,means the url must be external links.
   //if not external url,can be blocked from here. no need to continue execute the following code.
   //such as bres://upload.wikimedia....  etc .
-  if (!Utils::isExternalLink(req.url())) {
+  if (!Utils::isExternalLink(url)) {
     gdWarning( "Blocking element \"%s\" as built-in link ", req.url().toEncoded().data() );
     return new BlockedNetworkReply( this );
   }
@@ -205,17 +208,15 @@ QNetworkReply * ArticleNetworkAccessManager::createRequest( Operation op,
   {
     QByteArray referer = req.rawHeader( "Referer" );
 
-    //GD_DPRINTF( "Referer: %s\n", referer.data() );
-
     QUrl refererUrl = QUrl::fromEncoded( referer );
 
     //GD_DPRINTF( "Considering %s vs %s\n", getHostBase( req.url() ).toUtf8().data(),
     //        getHostBase( refererUrl ).toUtf8().data() );
 
-    if ( !req.url().host().endsWith( refererUrl.host() ) &&
-         getHostBase( req.url() ) != getHostBase( refererUrl ) && !req.url().scheme().startsWith("data") )
+    if ( !url.host().endsWith( refererUrl.host() ) &&
+         getHostBase( url ) != getHostBase( refererUrl ) && !url.scheme().startsWith("data") )
     {
-      gdWarning( "Blocking element \"%s\" due to not same domain", req.url().toEncoded().data() );
+      gdWarning( "Blocking element \"%s\" due to not same domain", url.toEncoded().data() );
 
       return new BlockedNetworkReply( this );
     }
@@ -241,15 +242,17 @@ QNetworkReply * ArticleNetworkAccessManager::createRequest( Operation op,
   }
 
   // spoof User-Agent
-  QNetworkRequest newReq(req);
-  if ( hideGoldenDictHeader && req.url().scheme().startsWith("http", Qt::CaseInsensitive))
+  QNetworkRequest newReq;
+  newReq.setUrl(url);
+  newReq.setAttribute( QNetworkRequest::RedirectPolicyAttribute, QNetworkRequest::NoLessSafeRedirectPolicy );
+  if ( hideGoldenDictHeader && url.scheme().startsWith("http", Qt::CaseInsensitive))
   {
     newReq.setRawHeader("User-Agent", req.rawHeader("User-Agent").replace(qApp->applicationName().toUtf8(), ""));
   }
 
   QNetworkReply *  reply = QNetworkAccessManager::createRequest( op, newReq, outgoingData );
 
-  if( req.url().scheme() == "https")
+  if( url.scheme() == "https")
   {
 #ifndef QT_NO_OPENSSL
     connect( reply, SIGNAL( sslErrors( QList< QSslError > ) ),
@@ -521,15 +524,11 @@ LocalSchemeHandler::LocalSchemeHandler(ArticleNetworkAccessManager& articleNetMg
 
 void LocalSchemeHandler::requestStarted(QWebEngineUrlRequestJob *requestJob)
 {
-    QUrl url = requestJob->requestUrl();
+  QUrl url = requestJob->requestUrl();
+  QNetworkRequest request;
+  request.setUrl( url );
 
-    QNetworkRequest request;
-    request.setUrl( url );
-
-    QNetworkReply* reply=this->mManager.createRequest(QNetworkAccessManager::GetOperation,request);
-    connect(reply,&QNetworkReply::finished,requestJob,[=](){
-        requestJob->reply("text/html",reply);
-    });
-    connect(requestJob, &QObject::destroyed, reply, &QObject::deleteLater);
+  QNetworkReply * reply = this->mManager.createRequest( QNetworkAccessManager::GetOperation, request );
+  connect( reply, &QNetworkReply::finished, requestJob, [ = ]() { requestJob->reply( "text/html", reply ); } );
+  connect( requestJob, &QObject::destroyed, reply, &QObject::deleteLater );
 }
-

article_netmgr.hh

@@ -66,7 +66,7 @@ public slots:
   // Own AllowFrameReply slots
   void applyMetaData();
   void applyError( QNetworkReply::NetworkError code );
-  void readDataFromBase();
+//  void readDataFromBase();
 
   // Redirect QNetworkReply slots
   virtual void abort()
@@ -214,5 +214,6 @@ protected:
 
 private:
     ArticleNetworkAccessManager& mManager;
+  QNetworkAccessManager mgr;
 };
 #endif

articleview.cc

@@ -347,7 +347,7 @@ ArticleView::ArticleView( QWidget * parent, ArticleNetworkAccessManager & nm, Au
   // Variable name for store current selection range
   rangeVarName = QString( "sr_%1" ).arg( QString::number( (quint64)this, 16 ) );
 
-  connect(GlobalBroadcaster::instance(), SIGNAL( emitDictIds(ActiveDictIds)), this,
+  connect(GlobalBroadcaster::instance(), SIGNAL( dictionaryChanges(ActiveDictIds)), this,
           SLOT(setActiveDictIds(ActiveDictIds)));
 
   channel = new QWebChannel(ui.definition->page());

globalbroadcaster.cpp

@@ -19,4 +19,11 @@ Config::Preferences * GlobalBroadcaster::getPreference()
   return preference;
 }
 
+void GlobalBroadcaster::addWhitelist(QString url){
+  whitelist.push_back(url);
+}
+
+bool GlobalBroadcaster::existedInWhitelist(QString url){
+  return std::find(whitelist.begin(), whitelist.end(), url) != whitelist.end();
+}
 // namespace global

globalbroadcaster.h

@@ -2,6 +2,7 @@
 #define GLOBAL_GLOBALBROADCASTER_H
 
 #include <QObject>
+#include <vector>
 #include "config.hh"
 
 struct ActiveDictIds
@@ -15,14 +16,16 @@ class GlobalBroadcaster : public QObject
   Q_OBJECT
 private:
   Config::Preferences * preference;
-
+  std::vector<QString> whitelist;
 public:
   void setPreference( Config::Preferences * _pre );
   Config::Preferences * getPreference();
   GlobalBroadcaster( QObject * parent = nullptr );
+  void addWhitelist(QString host);
+  bool existedInWhitelist(QString host);
   static GlobalBroadcaster * instance();
 signals:
-  void emitDictIds( ActiveDictIds ad );
+  void dictionaryChanges( ActiveDictIds ad );
 };
 
 #endif // GLOBAL_GLOBALBROADCASTER_H

goldendict.pro

@@ -226,6 +226,7 @@ DEFINES += PROGRAM_VERSION=\\\"$$VERSION\\\"
 HEADERS += folding.hh \
     article_inspect.h \
     globalbroadcaster.h \
+    iframeschemehandler.h \
     inc_case_folding.hh \
     inc_diacritic_folding.hh \
     mainwindow.hh \
@@ -366,6 +367,7 @@ FORMS += groups.ui \
 SOURCES += folding.cc \
     article_inspect.cpp \
     globalbroadcaster.cpp \
+    iframeschemehandler.cpp \
     main.cc \
     dictionary.cc \
     config.cc \

iframeschemehandler.cpp

@@ -0,0 +1,104 @@
+#include "iframeschemehandler.h"
+
+#include <QTextCodec>
+
+IframeSchemeHandler::IframeSchemeHandler(QObject * parent):QWebEngineUrlSchemeHandler(parent){
+
+}
+void IframeSchemeHandler::requestStarted(QWebEngineUrlRequestJob *requestJob)
+{
+  QUrl url = requestJob->requestUrl();
+
+  // website dictionary iframe url
+  url = QUrl( Utils::Url::queryItemValue( url, "url" ) );
+  QNetworkRequest request;
+  request.setUrl( url );
+
+  QNetworkReply * reply = mgr.get( request );
+  auto finishAction     = [ = ]() -> void
+  {
+    // Handle reply data
+
+    QByteArray replyData = reply->readAll();
+    QString articleString;
+
+    QTextCodec * codec = QTextCodec::codecForHtml( replyData );
+    if( codec )
+      articleString = codec->toUnicode( replyData );
+    else
+      articleString = QString::fromUtf8( replyData );
+
+    // Change links from relative to absolute
+
+    QString root = reply->url().scheme() + "://" + reply->url().host();
+    QString base = root + reply->url().path();
+    while( !base.isEmpty() && !base.endsWith( "/" ) )
+      base.chop( 1 );
+
+    QRegularExpression tags( "<\\s*(a|link|img|script)\\s+[^>]*(src|href)\\s*=\\s*['\"][^>]+>",
+                             QRegularExpression::CaseInsensitiveOption );
+    QRegularExpression links( "\\b(src|href)\\s*=\\s*(['\"])([^'\"]+['\"])",
+                              QRegularExpression::CaseInsensitiveOption );
+    int pos = 0;
+    QString articleNewString;
+    QRegularExpressionMatchIterator it = tags.globalMatch( articleString );
+    while( it.hasNext() )
+    {
+      QRegularExpressionMatch match = it.next();
+      articleNewString += articleString.mid( pos, match.capturedStart() - pos );
+      pos = match.capturedEnd();
+
+      QString tag = match.captured();
+
+      QRegularExpressionMatch match_links = links.match( tag );
+      if( !match_links.hasMatch() )
+      {
+        articleNewString += tag;
+        continue;
+      }
+
+      QString url = match_links.captured( 3 );
+
+      if( url.indexOf( ":/" ) >= 0 || url.indexOf( "data:" ) >= 0 || url.indexOf( "mailto:" ) >= 0 ||
+          url.startsWith( "#" ) || url.startsWith( "javascript:" ) )
+      {
+        // External link, anchor or base64-encoded data
+        articleNewString += tag;
+        continue;
+      }
+
+      QString newUrl = match_links.captured( 1 ) + "=" + match_links.captured( 2 );
+      if( url.startsWith( "//" ) )
+        newUrl += reply->url().scheme() + ":";
+      else if( url.startsWith( "/" ) )
+        newUrl += root;
+      else
+        newUrl += base;
+      newUrl += match_links.captured( 3 );
+
+      tag.replace( match_links.capturedStart(), match_links.capturedLength(), newUrl );
+      articleNewString += tag;
+    }
+    if( pos )
+    {
+      articleNewString += articleString.mid( pos );
+      articleString = articleNewString;
+      articleNewString.clear();
+    }
+
+    sptr< Dictionary::DataRequestInstant > response = new Dictionary::DataRequestInstant( true );
+
+    auto content = articleString.toStdString();
+    response->getData().resize( content.size() );
+    memcpy( &( response->getData().front() ), content.data(), content.size() );
+
+    auto contentType="text/html";
+    auto newReply = new ArticleResourceReply( this, request, response, contentType );
+
+    requestJob->reply( contentType, newReply );
+    connect( requestJob, &QObject::destroyed, newReply, &QObject::deleteLater );
+  };
+  connect( reply, &QNetworkReply::finished, requestJob, finishAction );
+
+  connect( requestJob, &QObject::destroyed, reply, &QObject::deleteLater );
+}

iframeschemehandler.h

@@ -0,0 +1,19 @@
+#ifndef IFRAMESCHEMEHANDLER_H
+#define IFRAMESCHEMEHANDLER_H
+
+#include"article_netmgr.hh"
+
+class IframeSchemeHandler : public QWebEngineUrlSchemeHandler
+{
+    Q_OBJECT
+public:
+    IframeSchemeHandler(QObject * parent=nullptr);
+    void requestStarted(QWebEngineUrlRequestJob *requestJob);
+
+protected:
+
+private:
+      QNetworkAccessManager mgr;
+};
+
+#endif // IFRAMESCHEMEHANDLER_H

main.cc

@@ -253,7 +253,7 @@ int main( int argc, char ** argv )
 
 #endif
 
-  QStringList localSchemes={"gdlookup","gdau","gico","qrcx","bres","bword","gdprg","gdvideo","gdpicture","gdtts"};
+  QStringList localSchemes={"gdlookup","gdau","gico","qrcx","bres","bword","gdprg","gdvideo","gdpicture","gdtts","ifr"};
 
   for (int i = 0; i < localSchemes.size(); ++i)
   {

mainwindow.cc

@@ -145,14 +145,19 @@ MainWindow::MainWindow( Config::Class & cfg_ ):
 
   GlobalBroadcaster::instance()->setPreference(&cfg.preferences);
 
-  localSchemeHandler = new LocalSchemeHandler(articleNetMgr);
-  QWebEngineProfile::defaultProfile()->installUrlSchemeHandler("gdlookup", localSchemeHandler);
-  QWebEngineProfile::defaultProfile()->installUrlSchemeHandler("bword", localSchemeHandler);
+  localSchemeHandler = new LocalSchemeHandler( articleNetMgr );
+  QWebEngineProfile::defaultProfile()->installUrlSchemeHandler( "gdlookup", localSchemeHandler );
+  QWebEngineProfile::defaultProfile()->installUrlSchemeHandler( "bword", localSchemeHandler );
 
-  QStringList localSchemes={"gdau","gico","qrcx","bres","gdprg","gdvideo","gdpicture","gdtts"};
-  resourceSchemeHandler = new ResourceSchemeHandler(articleNetMgr);
-  for(int i=0;i<localSchemes.size();i++){
-    QWebEngineProfile::defaultProfile()->installUrlSchemeHandler(localSchemes.at(i).toLatin1(), resourceSchemeHandler);
+  iframeSchemeHandler = new IframeSchemeHandler( this );
+  QWebEngineProfile::defaultProfile()->installUrlSchemeHandler( "ifr", iframeSchemeHandler );
+
+  QStringList localSchemes = { "gdau", "gico", "qrcx", "bres", "gdprg", "gdvideo", "gdpicture", "gdtts" };
+  resourceSchemeHandler    = new ResourceSchemeHandler( articleNetMgr );
+  for( int i = 0; i < localSchemes.size(); i++ )
+  {
+    QWebEngineProfile::defaultProfile()->installUrlSchemeHandler( localSchemes.at( i ).toLatin1(),
+                                                                  resourceSchemeHandler );
   }
 
   wuri = new WebUrlRequestInterceptor();

mainwindow.hh

@@ -34,6 +34,7 @@
 #include "hotkeywrapper.hh"
 #include "weburlrequestinterceptor.h"
 #include "resourceschemehandler.h"
+#include "iframeschemehandler.h"
 #ifdef HAVE_X11
 #include <fixx11h.h>
 #endif
@@ -176,8 +177,9 @@ private:
 
   QIcon starIcon, blueStarIcon;
 
-  LocalSchemeHandler *localSchemeHandler;
-  ResourceSchemeHandler *resourceSchemeHandler;
+  LocalSchemeHandler * localSchemeHandler;
+  IframeSchemeHandler * iframeSchemeHandler;
+  ResourceSchemeHandler * resourceSchemeHandler;
 
   /// Applies the qt's stylesheet, given the style's name.
   void applyQtStyleSheet( QString const & displayStyle, QString const & addonStyle );

website.cc

@@ -9,6 +9,7 @@
 #include <QDir>
 #include <QFileInfo>
 #include "gddebug.hh"
+#include "globalbroadcaster.h"
 
 #include <QRegularExpression>
 
@@ -371,8 +372,12 @@ sptr< DataRequest > WebSiteDictionary::getArticle( wstring const & str,
 
     string result = "<div class=\"website_padding\"></div>";
 
+    //permissive add url to global whitelist.
+    QUrl url(urlString);
+    GlobalBroadcaster::instance()->addWhitelist(url.host());
+
     result += string( "<iframe id=\"gdexpandframe-" ) + getId() +
-                      "\" src=\"" + urlString.data() +
+              "\" src=\"ifr://localhost?url=" +QUrl::toPercentEncoding(  urlString).data() +
                       "\" onmouseover=\"processIframeMouseOver('gdexpandframe-" + getId() + "');\" "
                       "onmouseout=\"processIframeMouseOut();\" "
                       "scrolling=\"no\" marginwidth=\"0\" marginheight=\"0\" "

weburlrequestinterceptor.cpp

@@ -1,6 +1,7 @@
 #include "weburlrequestinterceptor.h"
 #include <QDebug>
 #include "utils.hh"
+#include "globalbroadcaster.h"
 
 WebUrlRequestInterceptor::WebUrlRequestInterceptor(QObject *p)
   :QWebEngineUrlRequestInterceptor(p)
@@ -8,6 +9,14 @@ WebUrlRequestInterceptor::WebUrlRequestInterceptor(QObject *p)
 
 }
 void WebUrlRequestInterceptor::interceptRequest( QWebEngineUrlRequestInfo &info) {
+  if( Utils::isExternalLink( info.requestUrl() ) )
+  {
+    if(!GlobalBroadcaster::instance()-> existedInWhitelist(info.requestUrl().host()))
+    {
+      info.block( true );
+    }
+  }
+
   if (QWebEngineUrlRequestInfo::NavigationTypeLink == info.navigationType() && info.resourceType() == QWebEngineUrlRequestInfo::ResourceTypeMainFrame) {
     //workaround to fix devtool "Switch devtool to chinese" interface was blocked.
     if( info.requestUrl().scheme() == "devtools" )