"Document Type" field on .docx file isn't being saved to SharePoint

[aaronmorgan]

PnP.Core & PnP.Core.Auth, version: 1.8.38-nightly

I've got the most odd problem; a .docx file I created for testing (contains only some filler text) will not have one of it's fields set when a file in SharePoint.

In fact there's weird behaviour for Choices and Managed Metadata columns.

I upload a file like this:

IFile file = await documentsFolder.Files.AddAsync(
  name: parameter.Filename,
  content: fileStream,
  overwrite: true);

...and then set properties from a dictionary like this:

await file.LoadAsync(p => p.ListItemAllFields);

foreach (var property in parameter.Properties)
{
  file.ListItemAllFields[property.Key] = property.Value;
}

file.ListItemAllFields[SharePointFields.CONTENT_TYPEID] = parameter.ContentTypeId;
await file.ListItemAllFields.UpdateOverwriteVersionAsync();

If the file is a .docx the Document_x0020_Type column is not set, it's always null. Despite if I debug the code, the File's ListItemAllFields having the property set before the update is called.
Other columns and Choice fields are set correctly.

If the file is a PDF it works just fine.

I tried using some of the other update methods, e.g. UpdateAsync, and that made the problem worse because then none of the Choice fields were set either.

I've gone round and round on this and while other columns are working fine there's some weird interplay between the .docx file and Choice and Managed Metadata columns.

Have you seen or heard of anything like this?
What other information could I provide to help with understanding the cause?

offer_letter.docx

Thanks,
Aaron.

[jansenbe]

@aaronmorgan : can you try to load the relevant list information when you load ListItemAllFields (see https://pnp.github.io/pnpcore/using-the-sdk/files-intro.html#updatingreading-file-metadata for details). Internally we need to have the the list fields loaded to correctly identify the type of field. When possible this data is loaded automatically, but that's not always possible.

[jansenbe]

@aaronmorgan : did you experiment with adding a wait time after the file add and setting it's properties?

[aaronmorgan]

Ok, good idea; just now tried Thread.Sleep(10000) and same behaviour. Also using overwrite: false. The custom property still isn't set.

[aaronmorgan]

I've also upgraded to the latest nightly incase that helped.

[jansenbe]

@aaronmorgan : and if you leave out the content type update or first commit the content type update and then handle the fields coming with the content
type

[aaronmorgan]

Thanks Bert, that didn't affect the behaviour either.

I heard from one of our site testers that .xlsm files exhibit the same problem. This gels with the issue somehow being centered around newer Office documents.

We need to get to prod with this so I've worked around the problem by uploading the file with a .tmp extension to the filename, and simply remove it after setting the meta data and everything is fine.

Thanks for your help in trying to understand and resolve this for me. If I learn more about it I'll post back here.