Something weird with some pdf (valid for opening and printing) in lopdf 0.20.0

[stephaneworkspace]

Hello I found someting weird in some pdf file generated with PDFCreator 2.1 here is the pdfforge/PDFCreator: (github.com) but it’s the version 4.2.0, but for solving this problem, update the pdf creator version is not the solution because all old pdf have this problem. So… I linked to this issue 3 file generated with libre office → print → pdfcreator and files generated with libre office → generated pdf

After analysing in rust, for me, with the pdf creator, the format seems to be wrong, I solved the problem with a solution in this issue below.

I added a sw_debug and sw_pdf_creator in example of in main function at lopdf/examples/merge.rs

so… let sw_debug = true ; // diplay println! And let sw_pdf_creator = true ; // for solving pdf_creator

And :

    let mut j = 0;
    for mut document in documents {
        document.renumber_objects_with(max_id);

        max_id = document.max_id + 1;

        if sw_debug {
            for d in document.clone().get_pages().iter() {
                println!("{:?}", d);
            }
        }

        // Solving pdf_creator (min_value needed)
        let mut min_value = 0;
        if j > 0 {
            for d in document.clone().get_pages().iter() {
                if min_value == 0 {
                    min_value = d.1.0;
                    if sw_debug {
                        println!("debug: {:?}", d.1.0);
                    }
                }
                if d.1.0 < min_value {
                    min_value = d.1.0;
                    if sw_debug {
                        println!("debug: {:?}", d.1.0);
                    }
                }
            }
        }
        j += 1;
        if sw_debug {
            println!("j:{} {}",&j, &min_value);
        }
        if min_value > 0 {
            min_value -= 1;
        }

        documents_pages.extend(
            document
                .get_pages()
                .into_iter()
                .enumerate()
                .map(|(i, (_, object_id))| {
                    let object_id_mod = if i == 0 {
                        if sw_pdf_creator.clone() {
                            (min_value, object_id.1)
                        } else {
                            (object_id.0, object_id.1)
                        }
                    } else {
                        (object_id.0, object_id.1)
                    };
                    if sw_debug {
                        println!("mod: {:?} i: {}", object_id_mod, i);
                    }
                    (
                        object_id_mod,
                        document.get_object(object_id).unwrap().to_owned(),
                    )
                })
                .collect::<BTreeMap<ObjectId, Object>>(),
        );
        documents_objects.extend(document.objects);
    }

With merge of pdf_creator_a.pdf, pdf_creator_b.pdf and pdf_creator_c.pdf

I have in :

        if sw_debug {
            for d in document.clone().get_pages().iter() {
                println!("{:?}", d);
            }
        }

pdf_creator_a.pdf : (1, (21, 0)) (2, (1, 0)) (3, (6, 0)) (4, (11, 0))

pdf_creator_b.pdf : (1, (52, 0)) (2, (32, 0)) (3, (37, 0)) (4, (42, 0))

pdf_creator_c.pdf : (1, (83, 0)) (2, (63, 0)) (3, (68, 0)) (4, (73, 0))

With the mod trick :

                    if sw_debug {
                        println!("mod: {:?} i: {}", object_id_mod, i);
                    }

pdf_creator_a.pdf : mod: (0, 0) i: 0 mod: (1, 0) i: 1 mod: (6, 0) i: 2 mod: (11, 0) i: 3

pdf_creator_b.pdf : mod: (31, 0) i: 0 mod: (32, 0) i: 1 mod: (37, 0) i: 2 mod: (42, 0) i: 3

pdf_creator_c.pdf : mod: (62, 0) i: 0 mod: (63, 0) i: 1 mod: (68, 0) i: 2 mod: (73, 0) i: 3

No problem with pdf generated with libre office without pdf_creator (merge libre_office_a.pdf libre_office_b.pdf and libre_office_c.pdf) :

libre_office_a.pdf : (1, (1, 0)) (2, (4, 0)) (3, (7, 0)) (4, (10, 0))

libre_office_b.pdf : (1, (23, 0)) (2, (26, 0)) (3, (29, 0)) (4, (32, 0))

libre_office_c.pdf : (1, (45, 0)) (2, (48, 0)) (3, (51, 0)) (4, (54, 0))

libre_office_c.pdf libre_office_a.pdf libre_office_b.pdf pdf_creator_c.pdf pdf_creator_b.pdf pdf_creator_a.pdf

Thanks

[genusistimelord]

PDF_Creator does not re-correct the ID's before saving them which is what a proper PDF builder should be doing as Page 1 should not be ID 21 and page 2 being ID 1. The only way to fix this would be to add a ID swap feature which on load would see what ID's the pages are and swap them in the correct order. so it does not generate Madness when they are merged.

[Heinenen]

Maybe I'm misunderstanding something, but what is the problem? Having randomly assigned IDs isn't really a bad thing in PDFs, we have the Xref table to look them up in constant time anyway.