Pages are overwritten after merging PDFs with more than 50 pages about fpdi HOT 6 CLOSED

The logic you use results in a very dirty PDF which seems to trigger some limitations in PDF viewers. In e.g. Chrome or Foxit you will notice the issue at 42 pages, in pdf.js is no issue at all.
The problem is that you re-import imported pages and you do this again for each page you "add". The internal structure will hold several duplicates of each page:

...the tree goes much deeper.

So to fix this just use a single run to import all pages instead of adding each after each:

$filenames = [];

for ($i = 1; $i <= 100; $i++) {
    $dompdf = new Dompdf();
    $dompdf->setPaper('letter', 'landscape');
    $dompdf->loadHtml("<h1>Page $i</h1>");
    $dompdf->render();

    $output = $dompdf->output();

    $filename = "$basePath/page-$i.pdf";
    file_put_contents($filename, $output);
    $filenames[] = $filename;
}

$pdf = new Fpdi();

foreach($filenames AS $file) {
    $pageCount = $pdf->setSourceFile($file);

    for ($page = 1; $page <= $pageCount; $page++) {
        $pageId = $pdf->importPage($page);
        $templateSize = $pdf->getTemplatesize($pageId);
        $pdf->AddPage($templateSize['orientation'], $templateSize);
        $pdf->useImportedPage($pageId);
    }
}

$output = $pdf->Output('S');

$filename = "$basePath/merged.pdf";
file_put_contents($filename, $output);

You should understand that you cannot edit a PDF with FPDI but you import the pages of an existing one and place them onto new pages. If you do this recursively the internal structure will also grow appropriately and it looks like some PDF viewers have some limit in view to such structures. This may be an issue of a nesting level of graphic states - which was documented in PDF 1.7 but was removed in 2.0. Maybe it's simply a security check as such recursion looks sus.

Generally try to avoid importing of PDFs which were imported earlier but start with the original documents.
If you really want to append a PDF to another on a low level feel free to check out our SetaPDF-Merger component. It can be initiated with an existing document instance: https://manuals.setasign.com/setapdf-merger-manual/the-main-class/#index-2

from fpdi.

Hi @JanSlabon,

Thank you very much for this insightful reply. The code provided was just a test snippet, in the real application we can of course create the single PDFs before merging them. But I would still need to combine 100's of PDFs into a single document at some point. If I use your example code I quickly run into a "Failed to open stream: Too many open files" exception. Is there any way to close the files after adding them?

I will also take a look at the SetaPDF-Merger component, unfortunately with this project I'm pretty tight on paid solutions...

Thank you again!

from fpdi.

[...]"Failed to open stream: Too many open files"[...]

Long time not seen this error. Especially at only 100 files. You may increase this on OS level or split this at the max number (if you do this several times, you end in the same situation as you initially reported). But again the SetaPDF-Merger component got you covered: https://manuals.setasign.com/setapdf-merger-manual/performance-optimizations/#index-2

from fpdi.

Haha thanks, yeah the error did not happen around 100 files – more like 800 PDFs in. If there is no way to do this with an open source solution, I'll try to get the funds for the commercial licence. Thank you!

from fpdi.

You can try to hold the files in memory: https://manuals.setasign.com/fpdi-manual/v2/the-fpdi-class/#index-4
But internally FPDI makes use of streams - I'm not sure if this counts as file-handles, too.
If not it will, for sure, increase the memory usage which may trigger the next limitation (memory_limit).

Also for this the SetaPDF-Merger got you covered as you can work with an intermediate result to free memory: https://manuals.setasign.com/setapdf-merger-manual/performance-optimizations/#index-4

from fpdi.

Thank you again for your help. I was actually able to solve this with a simple shell_exec command and Ghostscript, this Gist was very helpful. The command only needs a few seconds for merging 800+ PDFs.

Closing this ...

from fpdi.