Question : Why does rotating page doesn't change mediaBox dimensions ?

[yodapotatofly]

I have a document with some pages being landscape, and the others being portrait.

I rotate all landscape pages to portrait (by checking mediaBox parameters and comparing x and y)
Then, I want to crop the pages in order to save only the top of the pages. But when i tried to crop a page that I have rotated, then everything goes sideways.
Actualy, the mediaBox parameters doesn't change between before and after rotating, while I think it should invert x and y sizes ?

from PyPDF2 import PdfFileWriter, PdfFileReader


def rotateLandscapePages(pdf):
    pdf_out = PdfFileWriter()
    numPages=pdf.getNumPages()


    for pageIndex in range(numPages):
        page = pdf.getPage(pageIndex)


        widht= page.mediaBox[2]
        height= page.mediaBox[3]
        print("\nPage",pageIndex+1)
        print("----------")
        print("widht :",widht, "; height:", height)
        print(page.mediaBox)
        print(page.mediaBox.lowerRight)
        print(page.mediaBox.lowerLeft)
        print(page.mediaBox.upperRight)
        print(page.mediaBox.upperLeft)


        if largeur > hauteur:
            print("Format paysage                     ",end=" ")
            print("Rotation à gauche")
            newPage = page.rotateCounterClockwise(90)

        else:
            print("Format portrait                     ",end=" ")
            print("Pas de rotation")
            newPage = page

        widht = newPage.mediaBox[2]
        height= newPage.mediaBox[3]
        print("\nPage",pageIndex+1)
        print("----------")
        print("widht :",widht, "; height :", height)
        print(newPage.mediaBox)
        print(newPage.mediaBox.lowerRight)
        print(newPage.mediaBox.lowerLeft)
        print(newPage.mediaBox.upperRight)
        print(newPage.mediaBox.upperLeft)

        cropTopHalf(newPage)
        pdf_out.addPage(newPage)

    return pdf_out

def cropTopHalf(page):
    page.mediaBox.lowerLeft = ( page.mediaBox.getLowerLeft_y(), page.mediaBox.getUpperRight_y()/2.1 )

if __name__ == '__main__':
    path = 'landscape_pdf_file.pdf'

    pdf = PdfFileReader(path)

    pdf = rotateLandscapePages(pdf)

    with open('out.pdf', 'wb') as fh:
        pdf.write(fh)

[thommiano]

Similarly, I am rotating pages (using page.RotateClockwise) that have annotations, but I've observed that the mediaBox and the annotation Rects do not have updated coordinates that reflect that page rotation. The rotation attribute also does not get updated.

If I write a rotate page to disk, then read it back in, the page looks rotated to the eye, but the annotations still have their original coordinates.

The expected behavior is that mediaBox, CropBox, Annotations, etc. have coordinates that reflect the current orientation/rotation of the page.

[maimaik05]

I am having this same issue, hoping to bring this back up if there is any input!

[maimaik05]

I am having this same issue, hoping to bring this back up if there is any input!

[pubpub-zz]

@maimaik05
many changes have to do to fix this.
a) can you confirm you are on pypdf 3.11
b) if you are still facing an issue, please fill up a issue with all requested data.

[pubpub-zz]

can you publish your pdf for review

[maimaik05]

Unfortunately, I can't share it, but I can say that it's a floorplan with AutoCAD Text as comments in the pdf.

[maimaik05]

Originally we were taking the pdf and using Adobe Acrobat to export the pdf as a jpg, then using PyPDF to find the location of the comments (annotations), by finding the center of the rect box of the annotation. It was working fine with that process. We figured out how to convert from pdf to jpg using python, but now the location of the annotation boxes appear to be rotated.

[pubpub-zz]

there is no logic in the document production between actual page orientation and "/Rotate" field. What you can try is to call PageObject.transfer_rotation_to_content() : this will modify the actual pageMedia and integration rotation within the content : this may fix your issue

[mohit1337]

PageObject.transfer_rotation_to_content() fixed the issue

[mohit1337]

This is still a problem. In below screenshots you can see, first one is original pdf of dimension 792x612

But even after rotating, the pdf is simply rotated to eye but dimensions are maintained

[pubpub-zz]

@mohit1337
can you please provide the PDF. if you prefer you can mail it privately to @MartinThoma

[stefan6419846]

The other comment (#704 (reply in thread)) has been added three minutes later, so I assume this could be resolved as well.

[pubpub-zz]

Not understood
I would like to see the PDF to analyse the behavior

[mohit1337]

@pubpub-zz, this is not pdf dependant, you can select any pdf of a rectangular dimension, load it and call .rotate(90) on the page and save it again and you shall be able to reproduce this.
Calling .transfer_rotation_to_content() on the page just before writing it fixed the issue though. This should either be default behaviour or made explicit in documents.