Impossible to run the squeezing of the PDF documents from Java due to JRE fatal issue in the cPDF library

[nickImbirev]

I'm playing with the library right now with one of my test experiments, but I keep bumping my head into this issue:

# A fatal error has been detected by the Java Runtime Environment:
#
#  SIGSEGV (0xb) at pc=0x00007f5349ca431a, pid=14, tid=121
#
# JRE version: OpenJDK Runtime Environment (17.0.2+8) (build 17.0.2+8-86)
# Java VM: OpenJDK 64-Bit Server VM (17.0.2+8-86, mixed mode, sharing, tiered, compressed oops, compressed class ptrs, g1 gc, linux-amd64)
# Problematic frame:
# C  [libcpdf.so+0x2f831a]  caml_alloc_minor_tables+0x7a

The code which I have integrated so far is very simple (following the examples from the project documentation):

        CompletableFuture.runAsync(
                        () -> {
                            Jcpdf pdfCompressor = new Jcpdf();
                            try {
                                pdfCompressor.startup();
                            } catch (Exception e) {
                                throw new RuntimeException(e);
                            }
                            try (var pdfContent = pdfCompressor.fromFile(file.toString(), emptyPassword)) {
                                pdfCompressor.squeezeInMemory(pdfContent);
                                pdfCompressor.toFile(pdfContent, file.toString(), false, false);
                            } catch (Exception e) {
                                throw new RuntimeException(e);
                            }
                        },
                        executorService)
                .join();

I would expect the library to work nicely in the concurrent environment, because unfortunately it is a blocker for me.

Please, let me know, if you need any further details

[nickImbirev]

example of the PDF document, which I used can be found here

[johnwhitington]

Jcpdf is not suitable for parallel or concurrent use. You would have to put your own lock around it, or use multiple java processes with such locks - each process would then have its own copy of libcpdf.dll / libjcpdf.dll.

Alternatively, you could use the cpdf command line tools, simply calling out to them.

(The command line tool is the original, upon which the APIs are based - which explains the single-threaded heritage.)

[johnwhitington]

In any event, don't call startup multiple times.

[nickImbirev]

Thank you for the quick reply @johnwhitington !

May I ask you, is this limitation being mentioned somewhere in the library doc?

[nickImbirev]

@johnwhitington a small question on top: would that be possible to use the cPDF version for JS environment with async/await? Does the library have any limitations for async engines?

I can consider re-writing the business logic in Node JS as well

[johnwhitington]

We can leave this open, as a not to fix the documentation for jcpdf for the next release.

My understanding of JavaScript async/await is that it is not actually parallel. Now, cpdf.js is pure javascript - so you should be ok, so long as calls to cpdf.js are, in the end, only happening one at a time, and in a sensible order. There is an example of a "web worker" style in the cpdf.js repository - though this example never queues up more than one job at a time.

Of course, you should expect cpdf.js to be a lot slower - squeeze is a complex operation - and you may need to configure node.js to use a bigger stack to deal with big PDFs.

[nickImbirev]

I see, thank you for the insights @johnwhitington !