How to ingest files with duplicate names but with different content?

[depuhitv]

I have a scenario where I am trying to ingest product documentation. I have the documentation stored in a folder tree for example

Administration 
|->1.0
|      |->configuration
|                 |-> server.conf
|-> 1.1
       |-> configuration
                 |-> server.conf

I originally moved the entire folder tree into the inputs folder assuming the tree would be ingested and the file_path would become metadata.
Unfortunately, it didn't seem to work and none of the files were ingested.

I couldn't find any documentation around how the bulk ingestion process works or if there is an alternate way of ingestion for an entire folder structure. Is there any guidance available?

Having the structure ingestion would be great, but I can understand if there arent many use cases for it and I can possibly use the breadcrumbs in the file instead for metadata.

But the other issue I have is that there are multiple files with the same name, but with different content but I have no way to of know of the conflict until after ingestion.

Is it possible to bypass the duplicate file name restriction globally?

[depuhitv]

I noticed the PR 1965 was the one where the recursive rglob was replaced by the plain glob, but there wasnt much of an explanation or reasoning mentioned or why it was mentioned as a fix.

The code does seem to work and also supports duplicate file names under different directory locations, so reverting to rglob would handle my ingestion scenario.

I've reverted that one change locally and ingestion is functioning in a test environment, but I haven't run a thorough test yet. I'm not sure if there was an architectural reason for removing the recursive capability. Could you shed some light on that?