This PR is a very over the top response to close #177
The issue there was that the previous handling of request responses was passing back just JSON data and error messages in a rather limited sort of way. That made it hard/hacky (string regexes on HTML text) to detect the reasons for a request failure. This PR creates new dataclasses to manage the resources and response handling for both the zenodo and metadata server use of requests. That removes some repetitious code as well as creating a cleaner and more flexible interface that includes explicit ok and status_code attributes. Changing that for Zenodo meant that we needed similar structures for the metadata server code.
That is all then updated in the entry_points.py, including some tidying of JSON data loading into one location.
Obviously, we don't yet have testing in place for this part of the package, but I've successfully published a dataset and new version of that dataset to the sandbox: https://sandbox.zenodo.org/records/91217
This PR is a very over the top response to close #177
The issue there was that the previous handling of
requestresponses was passing back just JSON data and error messages in a rather limited sort of way. That made it hard/hacky (string regexes on HTML text) to detect the reasons for a request failure. This PR creates new dataclasses to manage the resources and response handling for both the zenodo and metadata server use of requests. That removes some repetitious code as well as creating a cleaner and more flexible interface that includes explicitokandstatus_codeattributes. Changing that for Zenodo meant that we needed similar structures for the metadata server code.That is all then updated in the
entry_points.py, including some tidying of JSON data loading into one location.Obviously, we don't yet have testing in place for this part of the package, but I've successfully published a dataset and new version of that dataset to the sandbox: https://sandbox.zenodo.org/records/91217