I think this is a good feature request, as it is nice to get JSON dumped in a way that doesn't depend on the order of fields etc. (Not to mention sometimes it's just nice to have it sorted.)

I personally would welcome a PR adding this though I would defer to @samuelcolvin if he felt it was better not to.

Either way, adding the feature would need to happen in the implementation of model_dump_json in pydantic-core. @davidhewitt is it hard to get serde to dump JSON with keys in sorted order? It feels like something there might be an existing option for, or maybe not.

I'd be more than happy to contribute to this awesome project however I do not have any experience in Rust, yet 👀 .

In fact, there are other kwargs to JSON dumping that un-available such as separators.
Probably all captured in #6606 ?

Hi guys,

Would you mind giving us an update on this please?
I tried to look around with the aforementioned issues and open PR and could not make sense of it 😢

Thank you for your time :)

Why is this closed? @AqemiArnold will be be able to use sort_keys in model_dump_json() the same way we were able to use sort_keys in json() from v1? That is the issue here - a whole kwarg is unavailable.

Note: our schema requires sorted order in JSON string but it doesn't really matter in Python dicts.

I'm sorry I did not understand the message.

What are you (@antazoey) suggesting, and what should I do?

What are you (@antazoey) suggesting, and what should I do?

Here is the docs in the standard JSON library for dump: https://docs.python.org/3/library/json.html#basic-usage
Notice it has a kwarg sort_keys. It's doc is:

If sort_keys is true (default: False), then the output of dictionaries will be sorted by key.

This kwarg also works in Pydantic V1 in a model's .json() method.
This kwarg DOES NOT working on Pydantic v2 in a model's model_dump_json() method.

My feature request here is to add support for the sort_keys kwarg to Pydantic v2 the same way the standard JSON has and the same way Pydantic v1 has.

I have never contributed to Pydantic directly so I am not super aware of the technical challenges necessities in order to achieve this feature.

Hey everyone, sharing this snippet that's been a lifesaver for me 🚀:

instance_dict = instance.model_dump()
json_str = json.dumps(instance_dict, sort_keys=True)

Hope it helps anyone in the same boat!

It works but it is very slow for most of the use cases I have...

Hey everyone, sharing this snippet that's been a lifesaver for me 🚀:

instance_dict = instance.model_dump()
json_str = json.dumps(instance_dict, sort_keys=True)

Hope it helps anyone in the same boat!

Great snippet but this sadly doesn't work if you're using some of the custom PyDantic validators, like DirectoryPath for inst.

👍 to this and adding that it'd be nice to control this from the ConfigDict, e.g.

class MyModel(pydantic.BaseModel):
    model_config = pydantic.ConfigDict(sort_keys="asc")

This workaround is working for me:

https://gist.github.com/TweedBeetle/f56125de214318dc644fe5868bf5cd84

Just to note some technical details on this:
pydantic-core uses the preserve_order feature of serde_json:

• https://github.com/pydantic/pydantic-core/blob/6472887b3ad3e865e494b46efb66a71e87ceb1cf/Cargo.toml#L36
• https://github.com/serde-rs/json/blob/2ccb5b67ca34a5e53b42fa608d756e25e57f91a3/Cargo.toml#L56-L59

to "break" that ordering, I guess one would need to call https://docs.rs/serde_json/latest/serde_json/value/enum.Value.html#method.sort_all_objects or https://docs.rs/serde_json/latest/serde_json/map/struct.Map.html#method.sort_keys somewhere in the serialisation code

Is this still open?