Release a Qpdf without loading a new one

Armanta

Hello, I'm trying to unload a pdf file while keeping the object and without loading a file. My aim is to edit the pdf and reload it. I've already spent 5 hours on this problem, read tons of things and I can't do it.

I already check on https://forum.qt.io/topic/158080/how-can-i-release-a-pdf-document-for-use-again-after-it-has-been-used-with-qpdfview-in-pyqt/2. The only method that closes access is document.load but you need an other document (i dont want)

class EditorWidget(QFrame):
    def __init__(self, pdf_path: str, parent=None):
        super().__init__(parent=parent)
        self.setObjectName("EditorWidget")
        self.pdf_path = pdf_path

        self.layout = QVBoxLayout(self)
        self.layout.setContentsMargins(0, 0, 0, 0)
        
        self.commandBar = EditorCommandBar(self)
        self.layout.addWidget(self.commandBar)

        self.container = QWidget(self)
        self.container.setLayout(QVBoxLayout())
        self.container.layout().setContentsMargins(0, 0, 0, 0)
        
        # Initialisation du PDF viewer
        self.pdfView = CustomPdfView(self.container)
        self.pdfDocument = QPdfDocument(self)
        
        self.pdfDocument.load(self.pdf_path)
        self.pdfView.setDocument(self.pdfDocument)
        
        self.container.layout().addWidget(self.pdfView)
        self.layout.addWidget(self.container)
        
        # Créer le ScrollDelegate après avoir ajouté la vue
        self.scrollDelegate = SmoothScrollDelegate(self.pdfView)
        self.scrollDelegate.useAni = True
        self.scrollDelegate.eventFilter = lambda obj, event: False

        # Connecter les actions de zoom
        self.commandBar.zoomInAction.triggered.connect(self.pdfView.zoom_in)
        self.commandBar.zoomOutAction.triggered.connect(self.pdfView.zoom_out)
        self.commandBar.resetZoomAction.triggered.connect(self.pdfView.reset_zoom)

I would like to implement an unload_pdf method
Please don't tell me to make temporary files too

Armanta

I've finally found the solution, it was indeed the eventloop. The process will only be destroyed after the next loop.

match_manager.py

# Mettre à jour les variables dans le fichier Excel
        ExcelToPdfWorker.update_variables(excel_path, variables)

        # Connecter le signal avant de décharger le PDF
        match_widget.editor.pdf_unloaded.connect(
            lambda: self._convert_after_unload(match_widget, excel_path)
        )
        match_widget.editor.unload_pdf()

    def _convert_after_unload(self, match_widget, excel_path):
        """Appelé quand le PDF est vraiment déchargé."""
        self.excel_worker.excel_path = excel_path
        self.excel_worker.convert()
        # Une fois la conversion terminée, on peut recharger le PDF
        match_widget.editor.load_new_pdf(match_widget.pdf_path)

editor.py

def unload_pdf(self):
        """Ferme le PDF actuel et nettoie les références."""
        # Détacher la vue du document actuel
        self.pdfView.setDocument(None)
        
        # Fermer et supprimer l'ancien document
        if self.pdfDocument:
            old_document = self.pdfDocument
            # Créer un nouveau document vide avant de détruire l'ancien
            self.pdfDocument = QPdfDocument(self)
            # Connecter le signal destroyed à l'émission de notre signal
            old_document.destroyed.connect(self.pdf_unloaded.emit)
            old_document.close()
            old_document.deleteLater()
            
    def load_new_pdf(self, pdf_path):
        """Charge un nouveau fichier PDF."""
        # Charger le nouveau PDF
        self.pdfDocument.load(pdf_path)
        self.pdfView.setDocument(self.pdfDocument)
            
        print(f"Nouveau PDF chargé: {pdf_path}")
        
        return True

I'm well aware that my code sucks.

Here's the solution. As deleteLater() really destroy the PdfDocument and unload it, you have to wait until the end of a complete cycle and return to app.exec().
Then I put in a signal that will convert the file once the reference has been destroyed, since we know that this frees access

You should know that I think this is the only method without making temporary files

jsulm

@Armanta What about emptyPDF() method?

Armanta

@jsulm It sounds like emptyPDF() method doesn't exist.

https://doc.qt.io/qtforpython-6/search.html?q=emptyPDF&check_keywords=yes&area=default#

JonB

Did you try something like self.pdfDocument.close() followed by self.pdfDocument.load() again?
If not you might find that the self.pdfView.setDocument(self.pdfDocument) interferes. Try removing that to see if the QPdfDocument would successfully close() and then load() on its own. If so unset the document from the view before closing it and reattach the view afterwards.

Armanta

Just to explain how it works.

First i initialise the document and display it. Then I make some changes (in fact it's an Excel file that I convert to PDF).
So at first I have to display it. Then I have to unload it using an unload_pdf method, convert excel to pdf and display it again using a load_pdf method.

In my unload_pdf method I've already tried quite a few things, .close(), setDocument(None) etc... but despite this the pdf remains open. It's only by using the .load() method that the QPdfView gives up access.

And if I want to unload it using the load() method, you have to pass a file. I've already tried calling load without specifying a file (.load("")), it doesn't work and I don't want to pass a temporary file or anything.

@JonB If I remove self.pdfView.setDocument(self.pdfDocument)it will no longer be attached to my pdfView and will not be displayed.

alt text

JonB

@Armanta said in Release a Qpdf without loading a new one:

I've already tried calling load without specifying a file (.load("")),

I would have tried .load(None)?

If I remove self.pdfView.setDocument(self.pdfDocument)

Yes, but you could go

self.pdfView.setDocument(self.pdfDocument) # first time
# then whatever, till you want to reset the document
self.pdfView.setDocument(None)
# change what is is `self.pdfDocument` by reloading it or whatever
self.pdfView.setDocument(self.pdfDocument)

You could try above? I don't really know what is exactly that you want to happen when.

Armanta

Same problem, it crashed when going to self.pdfDocument.load(None)

# Third-party imports
from PySide6.QtCore import Qt
from PySide6.QtWidgets import QFrame, QVBoxLayout, QWidget, QHBoxLayout
from PySide6.QtPdfWidgets import QPdfView
from PySide6.QtPdf import QPdfDocument
from qfluentwidgets import (
    CommandBar, Action, FluentIcon as FIF, 
    SmoothScrollDelegate
)
class EditorWidget(QFrame):
    def __init__(self, pdf_path: str, parent=None):
        super().__init__(parent=parent)
        self.setObjectName("EditorWidget")
        self.pdf_path = pdf_path

        self.layout = QVBoxLayout(self)
        self.layout.setContentsMargins(0, 0, 0, 0)
        
        self.commandBar = EditorCommandBar(self)
        self.layout.addWidget(self.commandBar)

        self.container = QWidget(self)
        self.container.setLayout(QVBoxLayout())
        self.container.layout().setContentsMargins(0, 0, 0, 0)
        
        # Initialisation du PDF viewer
        self.pdfView = CustomPdfView(self.container)
        self.pdfDocument = QPdfDocument(self)
        
        self.pdfDocument.load(self.pdf_path)
        self.pdfView.setDocument(self.pdfDocument)
        
        self.container.layout().addWidget(self.pdfView)
        self.layout.addWidget(self.container)
        
        # Créer le ScrollDelegate après avoir ajouté la vue
        self.scrollDelegate = SmoothScrollDelegate(self.pdfView)
        self.scrollDelegate.useAni = True
        self.scrollDelegate.eventFilter = lambda obj, event: False

        # Connecter les actions de zoom
        self.commandBar.zoomInAction.triggered.connect(self.pdfView.zoom_in)
        self.commandBar.zoomOutAction.triggered.connect(self.pdfView.zoom_out)
        self.commandBar.resetZoomAction.triggered.connect(self.pdfView.reset_zoom)

    def unload_pdf(self):
        """Ferme le PDF actuel et nettoie les références."""
        self.pdfView.close()
        self.pdfDocument.close()
        self.pdfView.setDocument(None)
        self.pdfDocument.load(None)


    def load_new_pdf(self, pdf_path):
        """Charge un nouveau fichier PDF."""
        # Charger le nouveau PDF
        self.pdfDocument.load(pdf_path)
        self.pdfView.setDocument(self.pdfDocument)
            
        print(f"Nouveau PDF chargé: {pdf_path}")
        
        return True

JonB

@Armanta said in Release a Qpdf without loading a new one:

self.pdfDocument.load(None)

TBH I don't see why you need this at all. You have detached the view from the document and closed the document. What is wrong at this stage, what is the problem, what more do you want and why? If you can explain in one sentence with words of not more than three syllables that would be great :)

Armanta

The problem is that even after using the unload_pdf function, the document can be modified in read-write mode but cannot be deleted because it is still being used by the process. This then causes the conversion problem in Excel

Armanta

After a long debugging session I've just had a small piece of information: the problem is that Qt doesn't immediately release resources until the event loop has completed a full cycle

Armanta