This discussion has been locked.
You can no longer post new replies to this discussion. If you have a question you can start a new discussion

Problem with PDF conversion

Hi M-Files and community,

We have a nasty issue which causes us a lot of headache and problems. We produce PDFs through workflow using convert to PDF option. We use 22.10 version of M-Files server.

We have a problem with files containing long tables and images inside a document. These documents are not converted correctly to PDF. Quite often we miss values in metadata placeholders or sometimes documents are just incorrectly wrapped, parts of tables are missing, images are missing etc. When the same Word is converted with Adobe Pro, everything looks good and documents are formated correctly.

I've already had discussion with support and they have reproduced a problem with example document I've sent them.

The issue seems to be in Aspire library used for converting.

However, I've found following article talking exactly about our problem: forum.aspose.com/.../68408. if you read through, it seems that this bug is fixed in 2011! Does it mean that M-Files uses extremely old version of Aspose framework? It was told to us that the problem is 3rd party Aspose Library. It seems however that the problem is fixed.

Can you please advise what to do, how to mitigate this problem? I am just amazed that no one haven't raised this before as using tables with images is not unusual.

Thank you for sharing your experiences and thoughts.

Dejan

  • Hi again,

    I've looked into 22.10 installation and found that M-Files server uses 19.6.0.0 version of Aspose.Words for .NET: https://docs.aspose.com/words/net/aspose-words-for-net-19-6-release-notes/

    If you go through release notes, almost every new version has improvements related with rendering images.

    In 19.7 version there is interesting fix: "Improved table breaking in specific cases where table contains floating objects" which is one of cases we saw occurring. When we have images with tables, the table get wrapped around pages and some images (actually floating images disappear). 

    Wondering why M-Files is not updating Aspose libraries regularly? It seems that Aspose project is very active and fixes are provided regularly. 

    I see there 2 files Aspose.Words.dll and Aspose.Words.xml and also see it in many Apps. 

    Has anyone ever upgraded manually Aspose version due to similar issues in M-Files Server? I am slowly coming out of ideas but this issue is a major one.

    Dejan

  • To provide some updates:

    I tried to update M-Files server to use latest Aspose.Words for .NET. M-Files works stable until 20.9 version, it seems that afterwards the version M-Files is using are not valid anymore.

    However, this does not solve a problem, my example PDF still renders incorrectly, tables are wrapped over 2 pages, some images in tables are lost. Either M-Files is not using these libraries correctly or the issue is not solved until 20.9 version. I see many fixes in Aspose release notes, especially related to floating objects in Word documents such images and tables.

    Other issue, I am facing is that some complex documents (similar structure images and tables, the biggest one is 7 MB, so the size is not really a problem) are not getting metadata placeholders correctly updated in PDF. They are actually staying empty.

    The weird thing is that the Preview tab on Word documents as well is not showing any metadata in document. I've fixed preview to always create fresh PDF from Word so in normal cases it produces correct PDF. When I open Word, it contains all metadata placeholder correctly. When from this Word at the end of the workflow, PDF is produced, it contains no values on those metadata placeholders at all, no signature we maintained etc. I double checked all the metadata information and they are provided in the document but PDF just does not contain any of them.

    Preview mode clearly shows that it must be an issue somewhere in PDF rendering and usage of Aspose. We have many documents like this as they quite often contain tables and images in combination. It seems that M-Files rendering engine can not cope with complexity of the content (or perhaps Aspose can not do it even in 20.9 version?).

    If someone is interested, I can share a fake document explaining use case 1).

    Anyone experiencing anything similar?

    Best,

    Dejan

  • I recommend raising this with M-Files Support if you haven't yet done so and including some sample files where the issue can be seen in the support case. Support can then work with our product development in case any updates to third-party libraries are needed.