Downloading contents of folder

[GianniCarlo]

Context

When attempting to download the contents of a folder, we just use the current endpoint GET /v1/library?relativePath=$folderPath/&sign=true. The thing is, this response doesn't take into consideration the nested folders, as it's the same endpoint we use to list the library, so instead of getting the signed URLs of the contents of the nested folders, we get the URL for the s3 folder object.

Question

The previous scenario has some workarounds, as we can also request the type param and filter out client side the s3 folder URLs, to avoid downloading the s3 error response for folders, but this raises the question if it would be best to have a separate endpoint for fetching download URLs, so we can provide all the nested URLs, or if we even want to allow that (i.e. possibly downloading the entire library with just one action). I'm partial to just allowing downloads of one folder / level at a time, as it's rare that a user actually needs to have all their books on device, and not just their current listening series

[harrisi]

The /storage endpoint addition should make this easier, since /storage/:key will always download a file if it exists, and error otherwise. So GET /storage/someFolder and GET /storage/someFolder/ are both errors. To download all files from someFolder/, you need to GET /library?relativePath=someFolder/, which gives you all the top-level key names for that folder, and then GET /storage/${encodeURIComponent('someFolder/' + fileName)} for each file.

It's up to the client to decide if it should be recursive or not. I'd say, in general, it should be, but recommended to call GET /library/keys and see how many files have the prefix you're attempting to download, showing a warning message or something if it's a lot.