Transfer PDF annotations from broken.pdf to fixed.pdf

main.py

import PyPDF2

def transfer_annotations(src_pdf, dest_pdf, output_pdf):
    # Open the source and destination PDF files
    with open(src_pdf, 'rb') as src, open(dest_pdf, 'rb') as dest:
        src_reader = PyPDF2.PdfFileReader(src)
        dest_reader = PyPDF2.PdfFileReader(dest)
        pdf_writer = PyPDF2.PdfFileWriter()

        # Check the number of pages match in both PDFs
        if src_reader.numPages != dest_reader.numPages:
            print("Warning: The number of pages in the source and destination PDFs do not match.")
            return

        # Iterate over all the pages
        for page_num in range(src_reader.numPages):
            # Get the page from the destination PDF (since you want to retain its content)
            page = dest_reader.getPage(page_num)

            # Get the page from the source PDF to extract annotations
            src_page = src_reader.getPage(page_num)

            # Extract annotations from the source page
            annotations = src_page.get("/Annots")
            if annotations:
                # Resolve the IndirectObject to its actual value
                if isinstance(annotations, PyPDF2.generic.IndirectObject):
                    annotations = annotations.getObject()

                # If annotations are not present in the destination page, create an empty list
                if page.get("/Annots") is None:
                    page.__setitem__(PyPDF2.generic.NameObject("/Annots"), PyPDF2.generic.ArrayObject())

                # Append the annotations from the source page to the destination page
                page["/Annots"].extend(annotations)

            # Add the merged page to the PDF writer object
            pdf_writer.addPage(page)

        # Create the output PDF with merged annotations
        with open(output_pdf, 'wb') as out:
            pdf_writer.write(out)

    print(f"Annotations transferred. Output saved as {output_pdf}")

# Transfer annotations from broken.pdf to original.pdf
transfer_annotations("broken.pdf", "original.pdf", "output.pdf")

pip install PyPDF2==1.26.0