mirror of
https://github.com/xiaoyifang/goldendict-ng.git
synced 2024-11-24 16:54:08 +00:00
Merge remote-tracking branch 'origin/staged' into dev
This commit is contained in:
Xiao YiFang
commit
6a18eb9dac
|
|
@ -198,9 +198,6 @@ QNetworkReply * ArticleNetworkAccessManager::getArticleReply( QNetworkRequest co
|
|||
|
||||
QUrl refererUrl = QUrl::fromEncoded( referer );
|
||||
|
||||
//GD_DPRINTF( "Considering %s vs %s\n", getHostBase( req.url() ).toUtf8().data(),
|
||||
// getHostBase( refererUrl ).toUtf8().data() );
|
||||
|
||||
if ( !url.host().endsWith( refererUrl.host() ) &&
|
||||
getHostBaseFromUrl( url ) != getHostBaseFromUrl( refererUrl ) && !url.scheme().startsWith("data") )
|
||||
{
|
||||
|
|
@ -449,7 +446,7 @@ qint64 ArticleResourceReply::readData( char * out, qint64 maxSize )
|
|||
qint64 left = avail - alreadyRead;
|
||||
|
||||
qint64 toRead = maxSize < left ? maxSize : left;
|
||||
GD_DPRINTF( "====reading %d bytes\n", (int)toRead );
|
||||
GD_DPRINTF( "====reading %d bytes", (int)toRead );
|
||||
|
||||
try
|
||||
{
|
||||
|
|
@ -457,7 +454,7 @@ qint64 ArticleResourceReply::readData( char * out, qint64 maxSize )
|
|||
}
|
||||
catch( std::exception & e )
|
||||
{
|
||||
qWarning( "getDataSlice error: %s\n", e.what() );
|
||||
qWarning( "getDataSlice error: %s", e.what() );
|
||||
}
|
||||
|
||||
alreadyRead += toRead;
|
||||
|
|
|
|||
|
|
@ -402,6 +402,9 @@ void ArticleView::showDefinition( Config::InputPhrase const & phrase, unsigned g
|
|||
if ( scrollTo.size() )
|
||||
Utils::Url::addQueryItem( req, "scrollto", scrollTo );
|
||||
|
||||
if(delayedHighlightText.size())
|
||||
Utils::Url::addQueryItem( req, "regexp", delayedHighlightText );
|
||||
|
||||
Contexts::Iterator pos = contexts.find( "gdanchor" );
|
||||
if( pos != contexts.end() )
|
||||
{
|
||||
|
|
@ -579,6 +582,12 @@ void ArticleView::loadFinished( bool result )
|
|||
}
|
||||
if( Utils::Url::hasQueryItem( ui.definition->url(), "regexp" ) )
|
||||
highlightFTSResults();
|
||||
|
||||
if( !delayedHighlightText.isEmpty() )
|
||||
{
|
||||
// findText( delayedHighlightText, QWebEnginePage::FindCaseSensitively ,[](bool){});
|
||||
delayedHighlightText.clear();
|
||||
}
|
||||
}
|
||||
|
||||
void ArticleView::loadProgress(int ){
|
||||
|
|
@ -1592,6 +1601,11 @@ void ArticleView::setSelectionBySingleClick( bool set )
|
|||
ui.definition->setSelectionBySingleClick( set );
|
||||
}
|
||||
|
||||
void ArticleView::setDelayedHighlightText(QString const & text)
|
||||
{
|
||||
delayedHighlightText = text;
|
||||
}
|
||||
|
||||
void ArticleView::back()
|
||||
{
|
||||
// Don't allow navigating back to page 0, which is usually the initial
|
||||
|
|
@ -1712,6 +1726,7 @@ void ArticleView::contextMenuRequested( QPoint const & pos )
|
|||
QAction * sendWordToInputLineAction = 0;
|
||||
QAction * saveImageAction = 0;
|
||||
QAction * saveSoundAction = 0;
|
||||
QAction * saveBookmark = 0;
|
||||
|
||||
#if( QT_VERSION < QT_VERSION_CHECK( 6, 0, 0 ) )
|
||||
const QWebEngineContextMenuData * menuData = &(r->contextMenuData());
|
||||
|
|
@ -1832,6 +1847,13 @@ void ArticleView::contextMenuRequested( QPoint const & pos )
|
|||
}
|
||||
}
|
||||
|
||||
if(text.size())
|
||||
{
|
||||
// avoid too long in the menu ,use left 30 characters.
|
||||
saveBookmark = new QAction( tr( "Save &Bookmark \"%1...\"" ).arg( text.left( 30 ) ), &menu );
|
||||
menu.addAction( saveBookmark );
|
||||
}
|
||||
|
||||
// add anki menu
|
||||
if( !text.isEmpty() && cfg.preferences.ankiConnectServer.enabled )
|
||||
{
|
||||
|
|
@ -1931,7 +1953,11 @@ void ArticleView::contextMenuRequested( QPoint const & pos )
|
|||
QDesktopServices::openUrl( targetUrl );
|
||||
else
|
||||
if ( result == lookupSelection )
|
||||
showDefinition( selectedText, getGroup( ui.definition->url() ), getCurrentArticle() );
|
||||
showDefinition( text, getGroup( ui.definition->url() ), getCurrentArticle() );
|
||||
else if( result == saveBookmark )
|
||||
{
|
||||
emit saveBookmarkSignal( text.left( 60 ) );
|
||||
}
|
||||
else if( result == sendToAnkiAction )
|
||||
{
|
||||
sendToAnki( ui.definition->title(), ui.definition->selectedText() );
|
||||
|
|
@ -2333,7 +2359,11 @@ void ArticleView::performFindOperation( bool restart, bool backwards, bool check
|
|||
if ( backwards )
|
||||
f |= QWebEnginePage::FindBackward;
|
||||
|
||||
bool setMark = text.size() && !findText(text, f);
|
||||
findText( text,
|
||||
f,
|
||||
[ &text, this ]( bool match )
|
||||
{
|
||||
bool setMark = !text.isEmpty() && !match;
|
||||
|
||||
if( ui.searchText->property( "noResults" ).toBool() != setMark )
|
||||
{
|
||||
|
|
@ -2342,33 +2372,31 @@ void ArticleView::performFindOperation( bool restart, bool backwards, bool check
|
|||
// Reload stylesheet
|
||||
reloadStyleSheet();
|
||||
}
|
||||
} );
|
||||
}
|
||||
|
||||
bool ArticleView::findText(QString& text, const QWebEnginePage::FindFlags& f)
|
||||
void ArticleView::findText( QString & text,
|
||||
const QWebEnginePage::FindFlags & f,
|
||||
const std::function< void( bool match ) > & callback )
|
||||
{
|
||||
bool r;
|
||||
// turn async to sync invoke.
|
||||
QSharedPointer<QEventLoop> loop = QSharedPointer<QEventLoop>(new QEventLoop());
|
||||
QTimer::singleShot(1000, loop.data(), &QEventLoop::quit);
|
||||
#if( QT_VERSION >= QT_VERSION_CHECK( 6, 0, 0 ) )
|
||||
ui.definition->findText(text, f, [&](const QWebEngineFindTextResult& result)
|
||||
ui.definition->findText( text,
|
||||
f,
|
||||
[ callback ]( const QWebEngineFindTextResult & result )
|
||||
{
|
||||
if(loop->isRunning()){
|
||||
r = result.numberOfMatches()>0;
|
||||
loop->quit();
|
||||
} });
|
||||
auto r = result.numberOfMatches() > 0;
|
||||
if( callback )
|
||||
callback( r );
|
||||
} );
|
||||
#else
|
||||
ui.definition->findText(text, f, [&](bool result)
|
||||
ui.definition->findText( text,
|
||||
f,
|
||||
[ callback ]( bool result )
|
||||
{
|
||||
if(loop->isRunning()){
|
||||
r = result;
|
||||
loop->quit();
|
||||
} });
|
||||
if( callback )
|
||||
callback( result );
|
||||
} );
|
||||
#endif
|
||||
|
||||
|
||||
loop->exec();
|
||||
return r;
|
||||
}
|
||||
|
||||
void ArticleView::reloadStyleSheet()
|
||||
|
|
|
|||
|
|
@ -79,6 +79,8 @@ class ArticleView: public QFrame
|
|||
bool ftsSearchIsOpened, ftsSearchMatchCase;
|
||||
int ftsPosition;
|
||||
|
||||
QString delayedHighlightText;
|
||||
|
||||
void highlightFTSResults();
|
||||
void highlightAllFtsOccurences( QWebEnginePage::FindFlags flags );
|
||||
void performFtsFindOperation( bool backwards );
|
||||
|
|
@ -157,6 +159,8 @@ public:
|
|||
/// Called when preference changes
|
||||
void setSelectionBySingleClick( bool set );
|
||||
|
||||
void setDelayedHighlightText(QString const & text);
|
||||
|
||||
public slots:
|
||||
|
||||
/// Goes back in history
|
||||
|
|
@ -227,6 +231,10 @@ public:
|
|||
ResourceToSaveHandler * saveResource( const QUrl & url, const QString & fileName );
|
||||
ResourceToSaveHandler * saveResource( const QUrl & url, const QUrl & ref, const QString & fileName );
|
||||
|
||||
void findText( QString & text,
|
||||
const QWebEnginePage::FindFlags & f,
|
||||
const std::function< void( bool match ) > & callback = nullptr );
|
||||
|
||||
signals:
|
||||
|
||||
void iconChanged( ArticleView *, QIcon const & icon );
|
||||
|
|
@ -285,6 +293,8 @@ signals:
|
|||
|
||||
void inspectSignal(QWebEngineView * view);
|
||||
|
||||
void saveBookmarkSignal( const QString & bookmark );
|
||||
|
||||
public slots:
|
||||
|
||||
void on_searchPrevious_clicked();
|
||||
|
|
@ -391,7 +401,6 @@ private:
|
|||
|
||||
void performFindOperation( bool restart, bool backwards, bool checkHighlight = false );
|
||||
|
||||
bool findText(QString& text, const QWebEnginePage::FindFlags& f);
|
||||
|
||||
void reloadStyleSheet();
|
||||
|
||||
|
|
|
|||
|
|
@ -857,7 +857,7 @@ bool EpwingBook::getNextHeadword( EpwingHeadword & head )
|
|||
{
|
||||
EB_Position pos;
|
||||
|
||||
QRegularExpression badLinks( "#(v|n)\\d" );
|
||||
QRegularExpression badLinks( "#(v|n)\\d", QRegularExpression::UseUnicodePropertiesOption);
|
||||
|
||||
// At first we check references queue
|
||||
while( !LinksQueue.isEmpty() )
|
||||
|
|
|
|||
|
|
@ -71,8 +71,8 @@ bool parseSearchString( QString const & str, QStringList & indexWords,
|
|||
{
|
||||
searchWords.clear();
|
||||
indexWords.clear();
|
||||
QRegularExpression spacesRegExp( "\\W+" );
|
||||
QRegularExpression wordRegExp( QString( "\\w{" ) + QString::number( FTS::MinimumWordSize ) + ",}" );
|
||||
QRegularExpression spacesRegExp( "\\W+", QRegularExpression::UseUnicodePropertiesOption );
|
||||
QRegularExpression wordRegExp( QString( "\\w{" ) + QString::number( FTS::MinimumWordSize ) + ",}", QRegularExpression::UseUnicodePropertiesOption );
|
||||
QRegularExpression setsRegExp( "\\[[^\\]]+\\]", QRegularExpression::CaseInsensitiveOption );
|
||||
QRegularExpression regexRegExp( "\\\\[afnrtvdDwWsSbB]|\\\\x([0-9A-Fa-f]{4})|\\\\0([0-7]{3})", QRegularExpression::CaseInsensitiveOption);
|
||||
|
||||
|
|
|
|||
31
gddebug.cc
31
gddebug.cc
|
|
@ -11,30 +11,14 @@
|
|||
#endif
|
||||
|
||||
QFile * logFilePtr;
|
||||
static QTextCodec * utf8Codec;
|
||||
|
||||
void gdWarning( const char * msg, ... )
|
||||
{
|
||||
va_list ap;
|
||||
va_start( ap, msg );
|
||||
QTextCodec *localeCodec = 0;
|
||||
|
||||
if( logFilePtr && logFilePtr->isOpen() )
|
||||
{
|
||||
if( utf8Codec == 0 )
|
||||
utf8Codec = QTextCodec::codecForName( "UTF8" );
|
||||
|
||||
localeCodec = QTextCodec::codecForLocale();
|
||||
QTextCodec::setCodecForLocale( utf8Codec );
|
||||
}
|
||||
|
||||
qWarning() << QString().vasprintf( msg, ap );
|
||||
|
||||
if( logFilePtr && logFilePtr->isOpen() )
|
||||
{
|
||||
QTextCodec::setCodecForLocale( localeCodec );
|
||||
}
|
||||
|
||||
va_end( ap );
|
||||
}
|
||||
|
||||
|
|
@ -42,23 +26,8 @@ void gdDebug(const char *msg, ...)
|
|||
{
|
||||
va_list ap;
|
||||
va_start( ap, msg );
|
||||
// QTextCodec *localeCodec = 0;
|
||||
|
||||
// if( logFilePtr && logFilePtr->isOpen() )
|
||||
// {
|
||||
// if( utf8Codec == 0 )
|
||||
// utf8Codec = QTextCodec::codecForName( "UTF8" );
|
||||
|
||||
// localeCodec = QTextCodec::codecForLocale();
|
||||
// QTextCodec::setCodecForLocale( utf8Codec );
|
||||
// }
|
||||
|
||||
qDebug().noquote() << QString().vasprintf( msg, ap );
|
||||
|
||||
// if( logFilePtr && logFilePtr->isOpen() )
|
||||
// {
|
||||
// QTextCodec::setCodecForLocale( localeCodec );
|
||||
// }
|
||||
|
||||
va_end( ap );
|
||||
}
|
||||
|
|
|
|||
|
|
@ -12,7 +12,29 @@ system(git describe --tags --always --dirty): hasGit=1
|
|||
!isEmpty(hasGit){
|
||||
GIT_HASH=$$system(git rev-parse --short=8 HEAD )
|
||||
}
|
||||
system(echo $${VERSION}.$${GIT_HASH} > version.txt)
|
||||
|
||||
win32{
|
||||
# date /T output is locale aware.
|
||||
DD=$$system(date /T)
|
||||
DATE =$$replace(DD, / , )
|
||||
}
|
||||
else{
|
||||
DATE=$$system(date '+%y%m%d')
|
||||
}
|
||||
|
||||
system(echo $${VERSION}.$${GIT_HASH} on $${DATE} > version.txt)
|
||||
|
||||
!CONFIG( verbose_build_output ) {
|
||||
!win32|*-msvc* {
|
||||
# Reduce build log verbosity except for MinGW builds (mingw-make cannot
|
||||
# execute "@echo ..." commands inserted by qmake).
|
||||
CONFIG += silent
|
||||
}
|
||||
}
|
||||
|
||||
CONFIG( release, debug|release ) {
|
||||
DEFINES += NDEBUG
|
||||
}
|
||||
|
||||
# DEPENDPATH += . generators
|
||||
INCLUDEPATH += .
|
||||
|
|
|
|||
|
|
@ -64,7 +64,7 @@ bool IndexedZip::loadFile( uint32_t offset, vector< char > & data )
|
|||
|
||||
if ( !ZipFile::readLocalHeader( zip, header ) )
|
||||
{
|
||||
GD_DPRINTF( "Failed to load header\n" );
|
||||
GD_DPRINTF( "Failed to load header" );
|
||||
return false;
|
||||
}
|
||||
|
||||
|
|
@ -73,13 +73,13 @@ bool IndexedZip::loadFile( uint32_t offset, vector< char > & data )
|
|||
switch( header.compressionMethod )
|
||||
{
|
||||
case ZipFile::Uncompressed:
|
||||
GD_DPRINTF( "Uncompressed\n" );
|
||||
GD_DPRINTF( "Uncompressed" );
|
||||
data.resize( header.uncompressedSize );
|
||||
return (size_t) zip.read( &data.front(), data.size() ) == data.size();
|
||||
|
||||
case ZipFile::Deflated:
|
||||
{
|
||||
GD_DPRINTF( "Deflated\n" );
|
||||
GD_DPRINTF( "Deflated" );
|
||||
|
||||
// Now do the deflation
|
||||
|
||||
|
|
|
|||
|
|
@ -329,7 +329,12 @@
|
|||
<translation>引用的音频播放程序不存在。</translation>
|
||||
</message>
|
||||
<message>
|
||||
<location filename="../articleview.cc" line="1867"/>
|
||||
<location filename="../articleview.cc" line="1853"/>
|
||||
<source>Save &Bookmark "%1..."</source>
|
||||
<translation>保存为书签（&S）“%1...”</translation>
|
||||
</message>
|
||||
<message>
|
||||
<location filename="../articleview.cc" line="1861"/>
|
||||
<source>&Send "%1" to anki with selected text.</source>
|
||||
<translation>将“%1”发送到anki并附带选择的文本。</translation>
|
||||
</message>
|
||||
|
|
|
|||
|
|
@ -1690,6 +1690,7 @@ ArticleView * MainWindow::createNewTab( bool switchToIt,
|
|||
connect( view, SIGNAL( zoomIn()), this, SLOT( zoomin() ) );
|
||||
|
||||
connect( view, SIGNAL( zoomOut()), this, SLOT( zoomout() ) );
|
||||
connect( view, &ArticleView::saveBookmarkSignal, this, &MainWindow::addBookmarkToFavorite );
|
||||
|
||||
view->setSelectionBySingleClick( cfg.preferences.selectWordBySingleClick );
|
||||
|
||||
|
|
@ -3503,7 +3504,7 @@ void MainWindow::on_saveArticle_triggered()
|
|||
// MDict anchors
|
||||
QRegularExpression anchorLinkRe(
|
||||
"(<\\s*a\\s+[^>]*\\b(?:name|id)\\b\\s*=\\s*[\"']*g[0-9a-f]{32}_)([0-9a-f]+_)(?=[^\"'])",
|
||||
QRegularExpression::PatternOption::CaseInsensitiveOption );
|
||||
QRegularExpression::PatternOption::CaseInsensitiveOption|QRegularExpression::UseUnicodePropertiesOption );
|
||||
html.replace( anchorLinkRe, "\\1" );
|
||||
|
||||
if( complete )
|
||||
|
|
@ -4656,6 +4657,15 @@ void MainWindow::addWordToFavorites( QString const & word, unsigned groupId )
|
|||
ui.favoritesPaneWidget->addHeadword( folder, word );
|
||||
}
|
||||
|
||||
void MainWindow::addBookmarkToFavorite( QString const & text )
|
||||
{
|
||||
// get current tab word.
|
||||
QString word = unescapeTabHeader( ui.tabWidget->tabText( ui.tabWidget->currentIndex() ) );
|
||||
const auto bookmark = QString( "%1~~~%2" ).arg( word, text );
|
||||
|
||||
ui.favoritesPaneWidget->addHeadword( nullptr, bookmark );
|
||||
}
|
||||
|
||||
void MainWindow::addAllTabsToFavorites()
|
||||
{
|
||||
QString folder;
|
||||
|
|
@ -4728,8 +4738,22 @@ void MainWindow::headwordFromFavorites( QString const & headword,
|
|||
}
|
||||
|
||||
// Show headword without lost of focus on Favorites tree
|
||||
setTranslateBoxTextAndClearSuffix( headword, EscapeWildcards, DisablePopup );
|
||||
showTranslationFor(headword );
|
||||
// bookmark cases: the favorite item may like this "word~~~selectedtext"
|
||||
auto words = headword.split( "~~~" );
|
||||
|
||||
setTranslateBoxTextAndClearSuffix( words[0], EscapeWildcards, DisablePopup );
|
||||
|
||||
//must be a bookmark.
|
||||
if(words.size()>1)
|
||||
{
|
||||
auto view = getCurrentArticleView();
|
||||
if(view)
|
||||
{
|
||||
view->setDelayedHighlightText(words[1]);// findText( words[ 1 ], QWebEnginePage::FindCaseSensitively );
|
||||
}
|
||||
}
|
||||
|
||||
showTranslationFor( words[ 0 ] );
|
||||
}
|
||||
|
||||
#ifdef Q_OS_WIN32
|
||||
|
|
|
|||
|
|
@ -462,6 +462,8 @@ private slots:
|
|||
|
||||
void addWordToFavorites( QString const & word, unsigned groupId );
|
||||
|
||||
void addBookmarkToFavorite( QString const & text );
|
||||
|
||||
bool isWordPresentedInFavorites( QString const & word, unsigned groupId );
|
||||
|
||||
void sendWordToInputLine( QString const & word );
|
||||
|
|
|
|||
159
zim.cc
159
zim.cc
|
|
@ -78,7 +78,8 @@ enum CompressionType
|
|||
struct ZIM_header
|
||||
{
|
||||
quint32 magicNumber;
|
||||
quint32 version;
|
||||
quint16 majorVersion;
|
||||
quint16 minorVersion;
|
||||
quint8 uuid[ 16 ];
|
||||
quint32 articleCount;
|
||||
quint32 clusterCount;
|
||||
|
|
@ -125,7 +126,7 @@ __attribute__((packed))
|
|||
enum
|
||||
{
|
||||
Signature = 0x584D495A, // ZIMX on little-endian, XMIZ on big-endian
|
||||
CurrentFormatVersion = 1 + BtreeIndexing::FormatVersion + Folding::Version
|
||||
CurrentFormatVersion = 3 + BtreeIndexing::FormatVersion + Folding::Version
|
||||
};
|
||||
|
||||
struct IdxHeader
|
||||
|
|
@ -158,13 +159,15 @@ struct Cache
|
|||
quint32 clusterNumber;
|
||||
int stamp;
|
||||
int count, size;
|
||||
unsigned blobs_offset_size;
|
||||
|
||||
Cache() :
|
||||
data( 0 ),
|
||||
clusterNumber( 0 ),
|
||||
stamp( -1 ),
|
||||
count( 0 ),
|
||||
size( 0 )
|
||||
size( 0 ),
|
||||
blobs_offset_size( 0 )
|
||||
{}
|
||||
};
|
||||
|
||||
|
|
@ -184,13 +187,25 @@ public:
|
|||
}
|
||||
const ZIM_header & header() const
|
||||
{ return zimHeader; }
|
||||
string getClusterData( quint32 cluster_nom );
|
||||
|
||||
string getClusterData( quint32 cluster_nom, unsigned & blob_offset_size );
|
||||
|
||||
const QString getMimeType( quint16 nom )
|
||||
{ return mimeTypes.value( nom ); }
|
||||
|
||||
bool isArticleMime( quint16 mime_type )
|
||||
{ return getMimeType( mime_type ).startsWith( "text/html", Qt::CaseInsensitive )
|
||||
|| getMimeType( mime_type ).startsWith( "text/plain", Qt::CaseInsensitive ); }
|
||||
|
||||
|
||||
quint16 redirectedMimeType( RedirectEntry const & redEntry );
|
||||
|
||||
private:
|
||||
ZIM_header zimHeader;
|
||||
Cache cache[ CACHE_SIZE ];
|
||||
int stamp;
|
||||
QVector< QPair< quint64, quint32 > > clusterOffsets;
|
||||
QStringList mimeTypes;
|
||||
|
||||
void clearCache();
|
||||
};
|
||||
|
|
@ -291,10 +306,33 @@ bool ZimFile::open()
|
|||
|
||||
std::sort( clusterOffsets.begin(), clusterOffsets.end() );
|
||||
|
||||
// Read mime types
|
||||
|
||||
string type;
|
||||
char ch;
|
||||
|
||||
seek( zimHeader.mimeListPos );
|
||||
|
||||
for( ; ; )
|
||||
{
|
||||
type.clear();
|
||||
while( getChar( &ch ) )
|
||||
{
|
||||
if( ch == 0 )
|
||||
break;
|
||||
type.push_back( ch );
|
||||
}
|
||||
if( type.empty() )
|
||||
break;
|
||||
|
||||
QString s = QString::fromUtf8( type.c_str(), type.size() );
|
||||
mimeTypes.append( s );
|
||||
}
|
||||
|
||||
return true;
|
||||
}
|
||||
|
||||
string ZimFile::getClusterData( quint32 cluster_nom )
|
||||
string ZimFile::getClusterData( quint32 cluster_nom, unsigned & blobs_offset_size )
|
||||
{
|
||||
// Check cache
|
||||
int target = 0;
|
||||
|
|
@ -328,6 +366,7 @@ string ZimFile::getClusterData( quint32 cluster_nom )
|
|||
if( found )
|
||||
{
|
||||
// Cache hit
|
||||
blobs_offset_size = cache[ target ].blobs_offset_size;
|
||||
return string( cache[ target ].data, cache[ target ].count );
|
||||
}
|
||||
|
||||
|
|
@ -353,9 +392,11 @@ string ZimFile::getClusterData( quint32 cluster_nom )
|
|||
|
||||
seek( clusterOffsets.at( nom ).first );
|
||||
|
||||
char compressionType;
|
||||
if( !getChar( &compressionType ) )
|
||||
char compressionType, cluster_info;
|
||||
if( !getChar( &cluster_info ) )
|
||||
return string();
|
||||
compressionType = cluster_info & 0x0F;
|
||||
blobs_offset_size = cluster_info & 0x10 && zimHeader.majorVersion >= 6 ? 8 : 4;
|
||||
|
||||
string decompressedData;
|
||||
|
||||
|
|
@ -384,9 +425,16 @@ string ZimFile::getClusterData( quint32 cluster_nom )
|
|||
// Check BLOBs number in the cluster
|
||||
// We cache multi-element clusters only
|
||||
|
||||
quint32 firstOffset;
|
||||
quint32 firstOffset32;
|
||||
quint64 firstOffset;
|
||||
if( blobs_offset_size == 8 )
|
||||
memcpy( &firstOffset, decompressedData.data(), sizeof(firstOffset) );
|
||||
quint32 blobCount = ( firstOffset - 4 ) / 4;
|
||||
else
|
||||
{
|
||||
memcpy( &firstOffset32, decompressedData.data(), sizeof(firstOffset32) );
|
||||
firstOffset = firstOffset32;
|
||||
}
|
||||
quint32 blobCount = ( firstOffset - blobs_offset_size ) / blobs_offset_size;
|
||||
|
||||
if( blobCount > 1 )
|
||||
{
|
||||
|
|
@ -410,12 +458,52 @@ string ZimFile::getClusterData( quint32 cluster_nom )
|
|||
memcpy( cache[ target ].data, decompressedData.c_str(), size );
|
||||
cache[ target ].count = size;
|
||||
cache[ target ].clusterNumber = cluster_nom;
|
||||
cache[ target ].blobs_offset_size = blobs_offset_size;
|
||||
}
|
||||
}
|
||||
|
||||
return decompressedData;
|
||||
}
|
||||
|
||||
quint16 ZimFile::redirectedMimeType( RedirectEntry const & redEntry )
|
||||
{
|
||||
RedirectEntry current_entry = redEntry;
|
||||
quint64 current_pos = pos();
|
||||
quint16 mimetype = 0xFFFF;
|
||||
|
||||
for( ; ; )
|
||||
{
|
||||
quint32 current_nom = current_entry.redirectIndex;
|
||||
|
||||
seek( zimHeader.urlPtrPos + (quint64)current_nom * 8 );
|
||||
quint64 new_pos;
|
||||
if( read( reinterpret_cast< char * >( &new_pos ), sizeof(new_pos) ) != sizeof(new_pos) )
|
||||
break;
|
||||
|
||||
seek( new_pos );
|
||||
quint16 new_mimetype;
|
||||
if( read( reinterpret_cast< char * >( &new_mimetype ), sizeof(new_mimetype) ) != sizeof(new_mimetype) )
|
||||
break;
|
||||
|
||||
if( new_mimetype == 0xFFFF ) // Redirect to other article
|
||||
{
|
||||
if( read( reinterpret_cast< char * >( ¤t_entry ) + 2, sizeof( current_entry ) - 2 ) != sizeof( current_entry ) - 2 )
|
||||
break;
|
||||
if( current_nom == current_entry.redirectIndex )
|
||||
break;
|
||||
}
|
||||
else
|
||||
{
|
||||
mimetype = new_mimetype;
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
seek( current_pos );
|
||||
return mimetype;
|
||||
}
|
||||
|
||||
|
||||
// Some supporting functions
|
||||
|
||||
bool indexIsOldOrBad( string const & indexFile )
|
||||
|
|
@ -516,23 +604,42 @@ quint32 readArticle( ZimFile & file, quint32 articleNumber, string & result,
|
|||
|
||||
// Read cluster data
|
||||
|
||||
string decompressedData = file.getClusterData( artEntry.clusterNumber );
|
||||
unsigned offset_size = 0;
|
||||
string decompressedData = file.getClusterData( artEntry.clusterNumber, offset_size );
|
||||
if( decompressedData.empty() )
|
||||
break;
|
||||
|
||||
// Take article data from cluster
|
||||
|
||||
quint32 firstOffset;
|
||||
quint32 firstOffset32;
|
||||
quint64 firstOffset;
|
||||
|
||||
if( offset_size == 8 )
|
||||
memcpy( &firstOffset, decompressedData.data(), sizeof(firstOffset) );
|
||||
quint32 blobCount = ( firstOffset - 4 ) / 4;
|
||||
else
|
||||
{
|
||||
memcpy( &firstOffset32, decompressedData.data(), sizeof(firstOffset32) );
|
||||
firstOffset = firstOffset32;
|
||||
}
|
||||
quint32 blobCount = ( firstOffset - offset_size ) / offset_size;
|
||||
if( artEntry.blobNumber > blobCount )
|
||||
break;
|
||||
|
||||
quint32 size;
|
||||
if( offset_size == 8 )
|
||||
{
|
||||
quint64 offsets[ 2 ];
|
||||
memcpy( offsets, decompressedData.data() + artEntry.blobNumber * 8, sizeof(offsets) );
|
||||
size = offsets[ 1 ] - offsets[ 0 ];
|
||||
result.append( decompressedData, offsets[ 0 ], size );
|
||||
}
|
||||
else
|
||||
{
|
||||
quint32 offsets[ 2 ];
|
||||
memcpy( offsets, decompressedData.data() + artEntry.blobNumber * 4, sizeof(offsets) );
|
||||
quint32 size = offsets[ 1 ] - offsets[ 0 ];
|
||||
|
||||
size = offsets[ 1 ] - offsets[ 0 ];
|
||||
result.append( decompressedData, offsets[ 0 ], size );
|
||||
}
|
||||
|
||||
return articleNumber;
|
||||
}
|
||||
|
|
@ -1437,6 +1544,7 @@ vector< sptr< Dictionary::Class > > makeDictionaries(
|
|||
|
||||
df.open();
|
||||
ZIM_header const & zh = df.header();
|
||||
bool new_namespaces = ( zh.majorVersion >= 6 && zh.minorVersion >= 1 );
|
||||
|
||||
if( zh.magicNumber != 0x44D495A )
|
||||
throw exNotZimFile( i->c_str() );
|
||||
|
|
@ -1473,7 +1581,7 @@ vector< sptr< Dictionary::Class > > makeDictionaries(
|
|||
}
|
||||
|
||||
const quint64 * ptr;
|
||||
quint16 mimetype;
|
||||
quint16 mimetype, redirected_mime = 0xFFFF;
|
||||
ArticleEntry artEntry;
|
||||
RedirectEntry redEntry;
|
||||
string url, title;
|
||||
|
|
@ -1490,6 +1598,7 @@ vector< sptr< Dictionary::Class > > makeDictionaries(
|
|||
if( ret != sizeof(RedirectEntry) - 2 )
|
||||
throw exCantReadFile( i->c_str() );
|
||||
|
||||
redirected_mime = df.redirectedMimeType( redEntry );
|
||||
nameSpace = redEntry.nameSpace;
|
||||
}
|
||||
else
|
||||
|
|
@ -1501,7 +1610,7 @@ vector< sptr< Dictionary::Class > > makeDictionaries(
|
|||
|
||||
nameSpace = artEntry.nameSpace;
|
||||
|
||||
if( nameSpace == 'A' )
|
||||
if( ( nameSpace == 'A' || ( nameSpace == 'C' && new_namespaces ) ) && df.isArticleMime( mimetype ) )
|
||||
articleCount++;
|
||||
}
|
||||
|
||||
|
|
@ -1524,7 +1633,8 @@ vector< sptr< Dictionary::Class > > makeDictionaries(
|
|||
title.push_back( ch );
|
||||
}
|
||||
|
||||
if( nameSpace == 'A' )
|
||||
if( nameSpace == 'A' || ( nameSpace == 'C' && new_namespaces && ( df.isArticleMime( mimetype )
|
||||
|| ( mimetype == 0xFFFF && df.isArticleMime( redirected_mime ) ) ) ) )
|
||||
{
|
||||
wstring word;
|
||||
if( !title.empty() )
|
||||
|
|
@ -1532,6 +1642,9 @@ vector< sptr< Dictionary::Class > > makeDictionaries(
|
|||
else
|
||||
word = Utf8::decode( url );
|
||||
|
||||
if( df.isArticleMime( mimetype )
|
||||
|| ( mimetype == 0xFFFF && df.isArticleMime( redirected_mime ) ) )
|
||||
{
|
||||
if( maxHeadwordsToExpand && zh.articleCount >= maxHeadwordsToExpand )
|
||||
indexedWords.addSingleWord( word, n );
|
||||
else
|
||||
|
|
@ -1539,6 +1652,13 @@ vector< sptr< Dictionary::Class > > makeDictionaries(
|
|||
wordCount++;
|
||||
}
|
||||
else
|
||||
{
|
||||
url.insert( url.begin(), '/' );
|
||||
url.insert( url.begin(), nameSpace );
|
||||
indexedResources.addSingleWord( Utf8::decode( url ), n );
|
||||
}
|
||||
}
|
||||
else
|
||||
if( nameSpace == 'M' )
|
||||
{
|
||||
if( url.compare( "Title" ) == 0 )
|
||||
|
|
@ -1565,6 +1685,11 @@ vector< sptr< Dictionary::Class > > makeDictionaries(
|
|||
}
|
||||
}
|
||||
else
|
||||
if( nameSpace == 'X' )
|
||||
{
|
||||
continue;
|
||||
}
|
||||
else
|
||||
{
|
||||
url.insert( url.begin(), '/' );
|
||||
url.insert( url.begin(), nameSpace );
|
||||
|
|
|
|||
Loading…
Reference in a new issue