void RenderThread::run()
{
double factor = qBound(0.01, m_zoomFactor, 10.0);
if (!QFile::exists(m_pdfUrl)) return;
Poppler::Document* document = Poppler::Document::load(m_pdfUrl);
if (!document || document->isLocked()) {
delete document;
return;
}
// Access page of the PDF file
document->setRenderHint(Poppler::Document::Antialiasing, true);
document->setRenderHint(Poppler::Document::TextAntialiasing, true);
Poppler::Page* pdfPage = document->page(0); // Document starts at page 0
if (pdfPage == 0) {
return;
}
// Generate a QImage of the rendered page
QImage image = pdfPage->renderToImage(factor*200.0, factor*200.0);
emit previewReady(image);
delete pdfPage;
delete document;
}
void TestPassword::password3()
{
Poppler::Document *doc;
doc = Poppler::Document::load( QString::fromUtf8(TESTDATADIR "/unittestcases/PasswordEncrypted.pdf") );
QVERIFY( doc );
QVERIFY( doc->isLocked() );
QVERIFY( !doc->unlock( "", "password" ) );
QVERIFY( !doc->isLocked() );
delete doc;
}
void TestPassword::password2b()
{
Poppler::Document *doc;
doc = Poppler::Document::load(QString::fromUtf8(TESTDATADIR "/unittestcases/Gday garçon - owner.pdf") );
QVERIFY( doc );
QVERIFY( !doc->isLocked() );
QVERIFY( !doc->unlock( QString::fromUtf8("garçon").toLatin1(), "" ) );
QVERIFY( !doc->isLocked() );
delete doc;
}
// Index any PDFs that are attached. Basically it turns the PDF into text and adds it the same
// way as a note's body
void NoteIndexer::indexPdf(qint32 reslid) {
NSqlQuery sql(global.db);
if (reslid <= 0)
return;
QString file = global.fileManager.getDbaDirPath() + QString::number(reslid) +".pdf";
QString text = "";
Poppler::Document *doc = Poppler::Document::load(file);
if (doc == NULL || doc->isEncrypted() || doc->isLocked())
return;
for (int i=0; i<doc->numPages(); i++) {
QRectF rect;
text = text + doc->page(i)->text(rect) + QString(" ");
}
// Add the new content. it is basically a text version of the note with a weight of 100.
sql.prepare("Insert into SearchIndex (lid, weight, source, content) values (:lid, :weight, :source, :content)");
sql.bindValue(":lid", reslid);
sql.bindValue(":weight", 100);
sql.bindValue(":source", "recognition");
sql.bindValue(":content", text);
sql.exec();
}
// Index any PDFs that are attached. Basically it turns the PDF into text and adds it the same
// way as a note's body
void IndexRunner::indexPdf(qint32 lid, Resource &r) {
if (!keepRunning || pauseIndexing) {
indexTimer->start();
return;
}
ResourceTable rtable(db);
qint32 reslid = rtable.getLid(r.guid);
if (lid <= 0) {
indexTimer->start();
return;
}
QString file = global.fileManager.getDbaDirPath() + QString::number(reslid) +".pdf";
QString text = "";
Poppler::Document *doc = Poppler::Document::load(file);
if (doc == NULL || doc->isEncrypted() || doc->isLocked()) {
indexTimer->start();
return;
}
for (int i=0; keepRunning && !pauseIndexing && i<doc->numPages(); i++) {
QRectF rect;
text = text + doc->page(i)->text(rect) + QString(" ");
}
IndexRecord *rec = new IndexRecord();
rec->content = text;
rec->source = "recognition";
rec->weight = 100;
rec->lid = lid;
if (indexHash->contains(lid)) {
delete indexHash->value(lid);
indexHash->remove(lid);
}
indexHash->insert(lid, rec);
}
//TODO think about storing the documents instead of the filenames in a list
//Would avoid creation of the documents for each search but probably take up more memory
bool MainWindow::searchPDFPostResults(const QFileInfo &fi, QString searchString)
{
bool nHitsExceeded = false;
Poppler::Document *document = Poppler::Document::load(fi.absoluteFilePath());
if (!document || document->isLocked()) {
//Oops, document empty => error message
QMessageBox mbox;
mbox.setText(tr("Could not read the file \n'") + fi.fileName() +
tr("\ntherefore, possible matches might be missing"));
mbox.exec();
delete document;
return false;
}
//Get document title
QString title = document->info("Title");
//If no title specified => use filename instead
if(title == "")
title = fi.fileName();
QString shortTitle = QString(title);
if (shortTitle.length() > 16)
{
shortTitle = "..." + shortTitle.right(16);
}
ui->statusBar->showMessage("Durchsuche PDF: " + shortTitle);
qDebug() << "Searching pdf: " << title;
//Search the document for the searchString
int hitCounter = 0;
for (int page = 0; page < document->numPages(); page++) {
//Parameters 1,2,3,4 of the search function
//will contain the rectangle coordinates of where the text was found.
//Since we are not interested in these coordinates, we just pass a single variable
double coordinate = 0.0;
if (document->page(page)->search(searchString, coordinate,coordinate,coordinate,coordinate,
Poppler::Page::NextResult,
Poppler::Page::CaseInsensitive)) {
hitCounter++;
ui->statusBar->showMessage(tr("Searching in ") + shortTitle +
" - hit on page: " + QString::number(page));
//Yay, we found the search String on this page
//=> Add this page to the result list
addDocNPageToResultList(title, fi, page);
// TODO make this limit a variable
if(hitCounter >= maxHits)
{
QMessageBox mbox;
mbox.setText(tr("Search has already reached the maximum of ") + QString::number(maxHits)
+ tr(" hits please choose a more specific criterion"));
mbox.exec();
nHitsExceeded = true;
break;
}
}
}
//Don't forget to delete the document
delete document;
//Return whether the max number of hits was reached
return nHitsExceeded;
}
void guFolderInspector::extractIsbnsFromPdf(QString fileName, QList<QString> &ISBNList)
{
//PDFDoc *doc;
//doc = new PDFDoc()
//qDebug() << "file:" << fileName;
Poppler::Document* document;
document = Poppler::Document::load(fileName,0,0);
//Poppler::Document::
if (!document || document->isLocked()) {
// ... error message ....
delete document;
return;
}
// Paranoid safety check
if (!document) {
// ... error message ...
return;
}
Poppler::Page* pdfPage;
isbnMethods find;
// Access page of the PDF file
int numOfPages = document->numPages(); //количество страниц
int numOfSearchPages = 15;
//qDebug() << "num of pages " << numOfPages;
if(numOfPages < numOfSearchPages)
{
numOfSearchPages = numOfPages;
}
for ( int pageNumber = 0 ; pageNumber < numOfSearchPages ; pageNumber++)
{
pdfPage = document->page(pageNumber); // Document starts at page 0
if (!pdfPage) {
// ... error message ...
continue;
}
//QtMsgHandler qInstallMsgHandler ( QtMsgHandler handler )
QString pageContent;
pageContent = pdfPage->text(QRectF(QPointF(0,0),pdfPage->pageSizeF()));
find.findIsbns(pageContent, ISBNList);
// ... use image ...
//qDebug() << pageContent;
}
// after the usage, the page must be deleted
delete pdfPage;
//Finally, don't forget to destroy the document:
delete document;
}
int main( int argc, char **argv )
{
QApplication a( argc, argv ); // QApplication required!
QTime t;
t.start();
QDir directory( argv[1] );
foreach ( const QString &fileName, directory.entryList() ) {
if (fileName.endsWith("pdf") ) {
qDebug() << "Doing" << fileName.toLatin1().data() << ":";
Poppler::Document *doc = Poppler::Document::load( directory.canonicalPath()+"/"+fileName );
if (!doc) {
qWarning() << "doc not loaded";
} else if ( doc->isLocked() ) {
if (! doc->unlock( "", "password" ) ) {
qWarning() << "couldn't unlock document";
delete doc;
}
} else {
doc->pdfVersion();
doc->info("Title");
doc->info("Subject");
doc->info("Author");
doc->info("Keywords");
doc->info("Creator");
doc->info("Producer");
doc->date("CreationDate").toString();
doc->date("ModDate").toString();
doc->numPages();
doc->isLinearized();
doc->isEncrypted();
doc->okToPrint();
doc->okToCopy();
doc->okToChange();
doc->okToAddNotes();
doc->pageMode();
for( int index = 0; index < doc->numPages(); ++index ) {
Poppler::Page *page = doc->page( index );
QImage image = page->renderToImage();
page->pageSize();
page->orientation();
delete page;
std::cout << ".";
std::cout.flush();
}
std::cout << std::endl;
delete doc;
}
}
}
std::cout << "Elapsed time: " << (t.elapsed()/1000) << "seconds" << std::endl;
}
// Index any resources
void IndexRunner::indexRecognition(qint32 lid, Resource &r) {
if (!keepRunning || pauseIndexing) {
//indexTimer->start();
return;
}
// Add filename or source url to search index
if (r.attributes.isSet()) {
NSqlQuery sql(db);
ResourceAttributes a = r.attributes;
if (a.fileName.isSet()) {
sql.prepare("Insert into SearchIndex (lid, weight, source, content) values (:lid, :weight, :source, :content)");
sql.bindValue(":lid", lid);
sql.bindValue(":weight", 100);
sql.bindValue(":source", "recognition");
sql.bindValue(":content", QString(a.fileName));
sql.exec();
}
if (a.sourceURL.isSet()) {
sql.prepare("Insert into SearchIndex (lid, weight, source, content) values (:lid, :weight, :source, :content)");
sql.bindValue(":lid", lid);
sql.bindValue(":weight", 100);
sql.bindValue(":source", "recognition");
sql.bindValue(":content", QString(a.sourceURL));
sql.exec();
}
}
// Make sure we have something to look through.
Data recognition;
if (r.recognition.isSet())
recognition = r.recognition;
if (!recognition.body.isSet())
return;
QDomDocument doc;
QString emsg;
doc.setContent(recognition.body, &emsg);
// look for text tags
QDomNodeList anchors = doc.documentElement().elementsByTagName("t");
#if QT_VERSION < 0x050000
for (unsigned int i=0; keepRunning && !pauseIndexing && i<anchors.length(); i++) {
#else
for (int i=0; keepRunning && !pauseIndexing && i<anchors.length(); i++) {
#endif
QApplication::processEvents();
QDomElement enmedia = anchors.at(i).toElement();
QString weight = enmedia.attribute("w");
QString text = enmedia.text();
if (text != "") {
IndexRecord *rec = new IndexRecord();
rec->weight = weight.toInt();
rec->lid = lid;
rec->content = text;
rec->source = "recognition";
if (indexHash->contains(lid)) {
delete indexHash->value(lid);
indexHash->remove(lid);
}
indexHash->insert(lid, rec);
}
}
}
// Index any PDFs that are attached. Basically it turns the PDF into text and adds it the same
// way as a note's body
void IndexRunner::indexPdf(qint32 lid, Resource &r) {
if (!global.indexPDFLocally)
return;
if (!keepRunning || pauseIndexing) {
//indexTimer->start();
return;
}
ResourceTable rtable(db);
qint32 reslid = rtable.getLid(r.guid);
if (lid <= 0) {
//indexTimer->start();
return;
}
QString file = global.fileManager.getDbaDirPath() + QString::number(reslid) +".pdf";
QString text = "";
Poppler::Document *doc = Poppler::Document::load(file);
if (doc == nullptr || doc->isEncrypted() || doc->isLocked()) {
//indexTimer->start();
return;
}
for (int i=0; keepRunning && !pauseIndexing && i<doc->numPages(); i++) {
QRectF rect;
text = text + doc->page(i)->text(rect) + QString(" ");
}
IndexRecord *rec = new IndexRecord();
rec->content = text;
rec->source = "recognition";
rec->weight = 100;
rec->lid = lid;
if (indexHash->contains(lid)) {
delete indexHash->value(lid);
indexHash->remove(lid);
}
indexHash->insert(lid, rec);
}
// Index any files that are attached.
void IndexRunner::indexAttachment(qint32 lid, Resource &r) {
if (!officeFound)
return;
QLOG_DEBUG() << "indexing attachment to note " << lid;
if (!keepRunning || pauseIndexing) {
//indexTimer->start();
return;
}
ResourceTable rtable(db);
qint32 reslid = rtable.getLid(r.guid);
if (lid <= 0) {
//indexTimer->start();
return;
}
QLOG_DEBUG() << "Resource " << reslid;
QString extension = "";
ResourceAttributes attributes;
if (r.attributes.isSet())
attributes = r.attributes;
if (attributes.fileName.isSet()) {
extension = attributes.fileName;
int i = extension.indexOf(".");
if (i != -1)
extension = extension.mid(i);
}
if (extension != ".doc" && extension != ".xls" && extension != ".ppt" &&
extension != ".docx" && extension != ".xlsx" && extension != ".pptx" &&
extension != ".pps" && extension != ".pdf" && extension != ".odt" &&
extension != ".odf" && extension != ".ott" && extension != ".odm" &&
extension != ".html" && extension != ".txt" && extension != ".oth" &&
extension != ".ods" && extension != ".ots" && extension != ".odg" &&
extension != ".otg" && extension != ".odp" && extension != ".otp" &&
extension != ".odb" && extension != ".oxt" && extension != ".htm" &&
extension != ".docm")
return;
QString file = global.fileManager.getDbaDirPath() + QString::number(reslid) +extension;
QFile dataFile(file);
if (!dataFile.exists()) {
QDir dir(global.fileManager.getDbaDirPath());
QStringList filterList;
filterList.append(QString::number(lid)+".*");
QStringList list= dir.entryList(filterList, QDir::Files);
if (list.size() > 0) {
file = global.fileManager.getDbaDirPath()+list[0];
}
}
QString outDir = global.fileManager.getTmpDirPath();
QProcess sofficeProcess;
QString cmd = "soffice --headless --convert-to txt:\"Text\" --outdir "
+outDir + " "
+file;
sofficeProcess.start(cmd,
QIODevice::ReadWrite|QIODevice::Unbuffered);
QLOG_DEBUG() << "Starting soffice ";
sofficeProcess.waitForStarted();
QLOG_DEBUG() << "Waiting for completion";
sofficeProcess.waitForFinished();
int rc = sofficeProcess.exitCode();
QLOG_DEBUG() << "soffice Errors:" << sofficeProcess.readAllStandardError();
QLOG_DEBUG() << "soffice Output:" << sofficeProcess.readAllStandardOutput();
QLOG_DEBUG() << "return code:" << rc;
if (rc == 255) {
QLOG_ERROR() << "soffice not found. Disabling attachment indexing.";
this->officeFound = false;
return;
}
QFile txtFile(outDir+QString::number(reslid) +".txt");
if (txtFile.open(QIODevice::ReadOnly)) {
QString text;
text = txtFile.readAll();
NSqlQuery sql(db);
db->lockForWrite();
sql.prepare("Insert into SearchIndex (lid, weight, source, content) values (:lid, :weight, 'recognition', :content)");
sql.bindValue(":lid", lid);
sql.bindValue(":weight", 100);
text = global.normalizeTermForSearchAndIndex(text);
sql.bindValue(":content", text);
QLOG_DEBUG() << "Adding note resource to index DB";
sql.exec();
db->unlock();
txtFile.close();
}
QDir dir;
dir.remove(outDir+QString::number(reslid) +".txt");
}
void IndexRunner::flushCache() {
if (indexHash->size() <= 0)
return;
QDateTime start = QDateTime::currentDateTimeUtc();
NSqlQuery sql(db);
db->lockForWrite();
sql.exec("begin");
QHash<qint32, IndexRecord*>::iterator i;
// Start adding words to the index. Every 200 sql insertions we do a commit
int commitCount = 200;
for (i=indexHash->begin(); keepRunning && !pauseIndexing && i!=indexHash->end(); ++i) {
qint32 lid = i.key();
IndexRecord *rec = i.value();
qint32 weight = rec->weight;
QString source = rec->source;
QString content = rec->content;
delete rec;
// Delete any old content
sql.prepare("Delete from SearchIndex where lid=:lid and source=:source");
sql.bindValue(":lid", lid);
sql.bindValue(":source", source);
sql.exec();
// Add the new content. it is basically a text version of the note with a weight of 100.
sql.prepare("Insert into SearchIndex (lid, weight, source, content) values (:lid, :weight, :source, :content)");
sql.bindValue(":lid", lid);
sql.bindValue(":weight", weight);
sql.bindValue(":source", source);
content = global.normalizeTermForSearchAndIndex(content);
sql.bindValue(":content", content);
sql.exec();
commitCount--;
if (commitCount <= 0) {
sql.exec("commit");
commitCount = 200;
}
}
indexHash->clear();
sql.exec("commit");
sql.finish();
db->unlock();
QDateTime finish = QDateTime::currentDateTimeUtc();
QLOG_DEBUG() << "Index Cache Flush Complete: " <<
finish.toMSecsSinceEpoch() - start.toMSecsSinceEpoch()
<< " milliseconds.";
}
void IndexRunner::busy(bool value, bool finished) {
iAmBusy=value;
emit(this->indexDone(finished));
}
int main( int argc, char **argv )
{
QApplication a( argc, argv ); // QApplication required!
if ( argc < 2 ||
(argc == 3 && strcmp(argv[2], "-arthur") != 0) ||
argc > 3)
{
// use argument as file name
qWarning() << "usage: test-poppler-qt4 filename [-arthur]";
exit(1);
}
Poppler::Document *doc = Poppler::Document::load(QFile::decodeName(argv[1]));
if (!doc)
{
qWarning() << "doc not loaded";
exit(1);
}
if (doc->isLocked())
{
qWarning() << "document locked (needs password)";
exit(0);
}
if (doc->numPages() <= 0)
{
delete doc;
qDebug() << "Doc has no pages";
return 0;
}
QString backendString;
if (argc == 3 && strcmp(argv[2], "-arthur") == 0)
{
backendString = "Arthur";
doc->setRenderBackend(Poppler::Document::ArthurBackend);
}
else
{
backendString = "Splash";
doc->setRenderBackend(Poppler::Document::SplashBackend);
}
doc->setRenderHint(Poppler::Document::Antialiasing, true);
doc->setRenderHint(Poppler::Document::TextAntialiasing, true);
for (int i = 0; i < doc->numPages(); ++i)
{
Poppler::Page *page = doc->page(i);
if (page) {
qDebug() << "Rendering page using" << backendString << "backend: " << i;
QTime t = QTime::currentTime();
QImage image = page->renderToImage();
qDebug() << "Rendering took" << t.msecsTo(QTime::currentTime()) << "msecs";
image.save(QString("test-rennder-to-file%1.ppm").arg(i));
delete page;
}
}
return 0;
}
void PdfViewer::loadDocument(const QString &file, PdfView::PositionHandling keepPosition)
{
//QTime t = QTime::currentTime();
#ifndef QT_NO_CURSOR
QApplication::setOverrideCursor(Qt::WaitCursor);
#endif // QT_NO_CURSOR
// TODO: close only when the new file fails to load
const QString tempFileName = file; // we must copy file in a new variable because otherwise closeDocument() empties file if file == m_file (which is the case in slotReload())
closeDocument();
bool isLoaded = m_pdfView->load(tempFileName);
if (!isLoaded) {
QPointer<QMessageBox> msgBox = new QMessageBox(QMessageBox::Critical,
tr("Open Error"), tr("Cannot open:\n") + tempFileName,
QMessageBox::Ok, this);
msgBox->exec();
delete msgBox;
m_fileOpenRecentAction->removeFile(tempFileName);
#ifndef QT_NO_CURSOR
QApplication::restoreOverrideCursor();
#endif // QT_NO_CURSOR
return;
}
Poppler::Document *doc = m_pdfView->document();
while (doc->isLocked()) {
bool ok = true;
QString password = QInputDialog::getText(this, tr("Document Password"),
tr("Please insert the password of the document:"),
QLineEdit::Password, QString(), &ok);
if (!ok) {
m_pdfView->close();
m_fileOpenRecentAction->removeFile(tempFileName);
#ifndef QT_NO_CURSOR
QApplication::restoreOverrideCursor();
#endif // QT_NO_CURSOR
return;
}
doc->unlock(password.toLatin1(), password.toLatin1());
}
m_file = QFileInfo(tempFileName).absoluteFilePath();
m_fileOpenRecentAction->addFile(m_file);
// remove previous file from m_watcher and add new file
if (m_watcher)
{
const QStringList files = m_watcher->files();
if (!files.isEmpty())
m_watcher->removePaths(files);
m_watcher->addPath(m_file);
}
//qCritical() << t.msecsTo(QTime::currentTime());
Q_FOREACH(DocumentObserver *obs, m_observers) {
if (keepPosition == PdfView::DontKeepPosition)
obs->documentLoaded();
obs->pageChanged(m_currentPage, keepPosition);
//qCritical() << t.msecsTo(QTime::currentTime());
}
// set window title
const QString docTitle = doc->info("Title");
setWindowTitle((docTitle.isEmpty() ? QFileInfo(m_file).fileName() : docTitle)
+ " - " + QCoreApplication::applicationName());
// enable actions
m_fileSaveCopyAction->setEnabled(true);
m_findAction->setEnabled(true);
m_findNextAction->setEnabled(true);
m_findPreviousAction->setEnabled(true);
m_showPresentationAction->setEnabled(true);
#ifndef QT_NO_CURSOR
QApplication::restoreOverrideCursor();
#endif // QT_NO_CURSOR
//qCritical() << "close and load document" << t.msecsTo(QTime::currentTime());
}
