allow annotate to override model's fields

[Jan Napieralski]

I want to modify the QuerySet field such that it is fetched from DB and its real field is redacted. For example fetching some employee data, but replacing all the emails and phone numbers to NULL's because of some security or role flag in the API resolver for this

Currently the .annotate() method doesn't allow for the overlap of output field with model field. I have seen, that the way to handle it is to call .values() excluding the overwritten field, but in the large models with many fields or queries with joins, its not realistically possible to know what fields will be used when doing this QuerySet manipulation. The .deffer() doesn't work neither

I propose some flag to the .annotate() for allowing to overlap and essentially overwrite the model field, bc currently the only way without evaluating it, is the .extra()

[David Sanders]

Hi there, thanks for the idea though I believe (iirc) the original reason this was prevented was to avoid accidental data loss when saving the model.

I can't find the original thread discussing this but I did write up about this here: ​https://github.com/shangxiao/stupid-django-tricks/tree/master/annotation_overwrite

If this hack is unsuitable for you then I'd encourage you to start a thread on the Django Forum, ​https://forum.djangoproject.com/, to discuss the issue (any feature requests & proposals must get approval via the forum first). It's unlikely to be accepted but there may be other folks with other workarounds that may be your cup of tea 😉

[Jan Napieralski]

Thank you very much for the response, it was perfect for my case and its more ORM like than .extra()