Blank page after merge.

[morkai]

I'm trying to add a short text to each page of an existing document. The input document I'm using for testing has 18 pages and after running it through the specified script two pages containing only text are blank.

I've got permission to share the first three pages of the document (one of the failing pages is the page number 3).

Environment

Which environment were you using when you encountered the problem?

$ python -m platform
Windows-10-10.0.22621-SP0

$ python -c "import pypdf;print(pypdf._debug_versions)"
pypdf==3.16.4, crypt_provider=('pycryptodome', '3.10.1'), PIL=9.4.0

Code + PDF

This is a minimal, complete example that shows the issue:

from pypdf import PdfWriter, PdfReader, Transformation

input_pdf = PdfReader(open("blank_merge.input.pdf", "rb"))
output_pdf = PdfWriter()

for page_no, input_page in enumerate(input_pdf.pages):
    print(page_no + 1)
    footer_pdf = PdfReader(open("blank_merge.footer.pdf", "rb"))
    footer_page = footer_pdf.pages[0]
    footer_page.add_transformation(Transformation().rotate(0).translate(tx=0, ty=0))
    input_page.merge_page(footer_page)
    output_pdf.add_page(input_page)

outputStream = open("blank_merge.output.pdf", "wb")
output_pdf.write(outputStream)
outputStream.close()

In the above script we are merging the footer_page into the input_page, but if we swap the order and merge the input_page into the footer_page everything works (but is noticably slower):

from pypdf import PdfWriter, PdfReader, Transformation

input_pdf = PdfReader(open("blank_merge.input.pdf", "rb"))
output_pdf = PdfWriter()

for page_no, input_page in enumerate(input_pdf.pages):
    print(page_no + 1)
    footer_pdf = PdfReader(open("blank_merge.footer.pdf", "rb"))
    footer_page = footer_pdf.pages[0]
    footer_page.add_transformation(Transformation().rotate(0).translate(tx=0, ty=0))
    footer_page.merge_page(input_page)
    output_pdf.add_page(footer_page)

outputStream = open("blank_merge.output.pdf", "wb")
output_pdf.write(outputStream)
outputStream.close()

blank_merge.input.pdf - the input PDF - notice that the 3rd page is all text
blank_merge.footer.pdf - this is created through Chrome's 'Print to PDF' feature (PDF created with the reportlab package also doesn't work)
blank_merge.output.pdf - the result of merging the footer into the input - notice that the 3rd page is blank
blank_merge.output_reversed.pdf - the result of merging the input into the footer - notice that the 3rd page has the original contents and the extra footer

[stefan6419846]

I am not completely sure why, but the following approach seems to work (id est using the clone_from version):

from pypdf import PdfReader, PdfWriter, Transformation


writer = PdfWriter(clone_from="blank_merge.input.pdf")

for page_no, page in enumerate(writer.pages):
    print(page_no + 1)
    footer_pdf = PdfReader("blank_merge.footer.pdf")
    footer_page = footer_pdf.pages[0]
    footer_page.add_transformation(Transformation().rotate(0).translate(tx=0, ty=0))
    page.merge_page(footer_page)

writer.write("blank_merge.output.pdf")

[bigatti]

@morkai Could you test the same PDF, but using pypdf version 3.10.0 ?

I have the same problem, after version 3.10.0 there is a problem.

Compare 3.10.0 to 3.11.0

[morkai]

@bigatti Yes, it works in 3.10.0 and breaks from 3.11.0.

[stefan6419846]

Still no solution, but it seems like omitting the second page avoids the final blank page.

The offending section is

pypdf/pypdf/_page.py

Lines 1122 to 1126 in 4f2cd34

	except AttributeError:
	# applies at least for page not in writer
	# as a backup solution, we put content as an object although not in accordance with pdf ref
	# this will be fixed with the _add_object
	self[NameObject(PG.CONTENTS)] = content

This generates a standalone page object (84 0 obj) which seems to have the correct content (id est the content of the original page 3), but is never referenced. The actually referenced page content is in xref 79, but cannot be decoded as it uses plaintext while declaring the /FlateDecode filter.

Generated broken file (without the irrelevant first page): result_3.11.0.pdf