Adds a configuration option to allow mixed alias/unaliased field name use when deserializing

[matthew-chambers-pushly]

Problem statement:

Our company uses mashumaro extensively and have several interconnected services that use the same shared data structures; however, in some places it is useful for the serialized payloads to be human readable and in other places having the keys be "minified" using aliases is more important. To prevent needing to duplicate the dataclasses around the codebase, we use a combination of dialects and the by_alias= option to control how the data is serialized, but when the data is deserialized in the current mashumaro build the fieldnames must match the alias.

Changes:

Adds an option called loose_deserialization to the BaseConfig to allow for mixed aliased/unaliased usage during deserialization.

[pep8speaks]

Hello @matthew-chambers-pushly! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found:

There are currently no PEP 8 issues detected in this Pull Request. Cheers! 🍻

Comment last updated at 2023-11-14 15:16:43 UTC

[Fatal1ty]

Hi @matthew-chambers-pushly

This is something that was on my radar! Thank you for bringing it in. However, there are some things worth paying attention to.

I think that loose_deserialization is too broad name for this feature. I'm sure there is a more accurate name for it. How about allow_deserialization_not_by_alias ?

[matthew-chambers-pushly]

Those mypy failures seem to be a regression introduced by mypy 1.7.

[Fatal1ty]

Those mypy failures seem to be a regression introduced by mypy 1.7.

I fixed it in master. Please, update your branch.

[Fatal1ty]

@matthew-chambers-pushly

I made some tweaks to tests and docs before the merge. Thank you for your contribution!

[matthew-chambers-pushly]

@matthew-chambers-pushly

I made some tweaks to tests and docs before the merge. Thank you for your contribution!

Thank you for the library, we've shown the testing and coverage as an example to younger devs.