Background

I finally convinced someone willing to share his full archival node 5868GiB database for free (which now requires to be built in ram and thus requires 100000$ worth of ram in order to be built but can be run on an ssd once done).

However he want to send it only through sending a single tar file over raw tcp using a rather slow (400Mps) connection for this task.
I m needing to get it on dropbox and as a result, he don’t want to use the https://www.dropbox.com/request/[my upload key here] allowing to upload files through a web browser without a dropbox account (it really annoyed him that I talked about using an other method or compressing the database to the point he is on the verge of changing his mind about sharing it).
Because on my side, dropbox allows using 10Tib of storage for free during 30 days and I didn’t receive the required ssd yet (so once received I will be able to download it using a faster speed).

The problem

So the idea right now is to receive it through the method he want and pipe it using command line to the command for uploading it.

There are many tools for sending a file to dropbox. But all those I found requires downloading the whole file first or bufferring it in ram (even for fuse filesystems). I don t have the ram not even the disk space for that.

Is there even a curl command for uploading the file to my account or using the original https://www.dropbox.com/request/[my upload key here] link?