Open kshepherd opened 3 weeks ago
(i should note, you can also just bypass the python obj and use the dict directly like bitstream.embedded['format']['mimetype'] -- it's hard to predict with the different ways and context we use libs like this, whether the "knowledge" of how the data looks should be in the client impl, or further up in the library
@alanorth This is one way of fixing #18 -- I am interested to hear if you think it's the right approach.
We can request linked HAL stuff to be embedded in the parent object using
?embed=...
in the REST API call. So this change allows a client to saybitstreams = d.get_bitstreams(bundle=my_bundle, embeds=['format'])
, and then in the resulting Bitstream objects you can get format out likeformat = BitstreamFormat(bitstream.embedded['format'])
without any additional API calls.The only thing I wasn't sure about is if we should handle that in the client lib instead of making the actual client do it... this way keeps things relatively simple in the library, we don't have to know anything in the bitstream model because it'll just copy all _embedded JSON into a dict. If we wanted to make that model more 'smart' we'd tell it what kind of embeds to expect, and we could parse and instantiate the BitstreamFormat objects before handing back in the get_bitstreams() return value.
Here is my example script to test it out: