V1 models for response and body

[chbndrhnns]

No description provided.

[chbndrhnns]

I am not sure if there is a better way to handle the case where there is a container of model instances.

[wu-clan]

isinstance()?

if isinstance(value, dict): ...

[chbndrhnns]

At least for Python 3.11+, I expect its zero-overhead exception handling to be a bit more efficient. In case that's neglectable, an if/else branch with isinstance might be more readable.

[mjspeck]

Any progress on this? Looks like it's been unchanged for almost a month.

[conglei]

Any update on this PR? I'm also blocked by the same issue.

[aviau]

This would really help large projects migrate to pydantic v2!

[YanSte]

Any update ?

[zouyi100]

Any update?

[jabbapen]

Any update?

[fraser-langton]

Any update on this? FastAPI still supports Pydantic v1 is a fundamentally incorrect statement while this remains open, and we will not upgrade to pydantic v2 or fastapi 0.100 while this remains open. Otherwise there is no path to a smooth transition to v2, the entire repo must be upgraded in one go, defeating the purpose behind pydantic.v1 usage.

https://fastapi.tiangolo.com/release-notes/#pydantic-v1

This version [0.100.0] of FastAPI still supports Pydantic v1. And although Pydantic v1 will be deprecated at some point, it will still be supported for a while.

This means that you can install the new Pydantic v2, and if something fails, you can install Pydantic v1 while you fix any problems you might have, but having the latest FastAPI.

[shanumas]

Any update ?

[vfazio]

Any update on this? FastAPI still supports Pydantic v1 is a fundamentally incorrect statement while this remains open, and we will not upgrade to pydantic v2 or fastapi 0.100 while this remains open. Otherwise there is no path to a smooth transition to v2, the entire repo must be upgraded in one go, defeating the purpose behind pydantic.v1 usage.

https://fastapi.tiangolo.com/release-notes/#pydantic-v1

This version [0.100.0] of FastAPI still supports Pydantic v1. And although Pydantic v1 will be deprecated at some point, it will still be supported for a while.
This means that you can install the new Pydantic v2, and if something fails, you can install Pydantic v1 while you fix any problems you might have, but having the latest FastAPI.

@tiangolo thoughts on this?

[phillipuniverse]

Looks like this will work for serialization/deserialization but needs additional work for schema generation:

../../../Library/Caches/pypoetry/virtualenvs/shipwell-common-python-zjZyLK7l-py3.11/lib/python3.11/site-packages/fastapi/openapi/utils.py:475: in get_openapi field_mapping, definitions = get_definitions( ../../../Library/Caches/pypoetry/virtualenvs/shipwell-common-python-zjZyLK7l-py3.11/lib/python3.11/site-packages/fastapi/_compat.py:230: in get_definitions field_mapping, definitions = schema_generator.generate_definitions( ../../../Library/Caches/pypoetry/virtualenvs/shipwell-common-python-zjZyLK7l-py3.11/lib/python3.11/site-packages/pydantic/json_schema.py:379: in generate_definitions self.generate_inner(schema) ../../../Library/Caches/pypoetry/virtualenvs/shipwell-common-python-zjZyLK7l-py3.11/lib/python3.11/site-packages/pydantic/json_schema.py:552: in generate_inner json_schema = current_handler(schema) ../../../Library/Caches/pypoetry/virtualenvs/shipwell-common-python-zjZyLK7l-py3.11/lib/python3.11/site-packages/pydantic/_internal/_schema_generation_shared.py:37: in __call__ return self.handler(core_schema) ../../../Library/Caches/pypoetry/virtualenvs/shipwell-common-python-zjZyLK7l-py3.11/lib/python3.11/site-packages/pydantic/json_schema.py:511: in handler_func json_schema = generate_for_schema_type(schema_or_field) ../../../Library/Caches/pypoetry/virtualenvs/shipwell-common-python-zjZyLK7l-py3.11/lib/python3.11/site-packages/pydantic/json_schema.py:1229: in chain_schema return self.generate_inner(schema['steps'][step_index]) ../../../Library/Caches/pypoetry/virtualenvs/shipwell-common-python-zjZyLK7l-py3.11/lib/python3.11/site-packages/pydantic/json_schema.py:552: in generate_inner json_schema = current_handler(schema) ../../../Library/Caches/pypoetry/virtualenvs/shipwell-common-python-zjZyLK7l-py3.11/lib/python3.11/site-packages/pydantic/_internal/_schema_generation_shared.py:37: in __call__ return self.handler(core_schema) ../../../Library/Caches/pypoetry/virtualenvs/shipwell-common-python-zjZyLK7l-py3.11/lib/python3.11/site-packages/pydantic/json_schema.py:511: in handler_func json_schema = generate_for_schema_type(schema_or_field) ../../../Library/Caches/pypoetry/virtualenvs/shipwell-common-python-zjZyLK7l-py3.11/lib/python3.11/site-packages/pydantic/json_schema.py:1014: in function_plain_schema return self.handle_invalid_for_json_schema( _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ self = <pydantic.json_schema.GenerateJsonSchema object at 0x113e3f0d0> schema = {'function': {'function': <bound method BaseModel.validate of <class 'tests.contrib.fastapi.test_shipwell_fastapi_extension.shipwell_fast_api.<locals>.Price'>>, 'type': 'with-info'}, 'metadata': {}, 'type': 'function-plain'} error_info = "core_schema.PlainValidatorFunctionSchema ({'type': 'with-info', 'function': <bound method BaseModel.validate of <class 'tests.contrib.fastapi.test_shipwell_fastapi_extension.shipwell_fast_api.<locals>.Price'>>})" def handle_invalid_for_json_schema(self, schema: CoreSchemaOrField, error_info: str) -> JsonSchemaValue: > raise PydanticInvalidForJsonSchema(f'Cannot generate a JsonSchema for {error_info}') E pydantic.errors.PydanticInvalidForJsonSchema: Cannot generate a JsonSchema for core_schema.PlainValidatorFunctionSchema ({'type': 'with-info', 'function': <bound method BaseModel.validate of <class 'tests.contrib.fastapi.test_shipwell_fastapi_extension.shipwell_fast_api.<locals>.Price'>>}) E E For further information visit https://errors.pydantic.dev/2.9/u/invalid-for-json-schema 

In this case the v2 schema generation is looking at the validate method on pydantic.v1.BaseModel and trying to generate a schema for it.

[YuriiMotov]

@chbndrhnns, thanks for you contribution!
Are you still ready to work on it?

Or, is there anybody who is willing to take this over and work on it?

[chbndrhnns]

For now, I updated the code to have passing tests.
Looking into schema generation now.

[chbndrhnns]

Schema generation seems a bit more tricky:

FastAPI now needs to handle both Pydantic v1 and v2 models simultaneously when running with Pydantic v2 installed. This is fundamentally different from the previous situation:

Before: FastAPI only needed to handle one version of Pydantic models (v1)
Now: FastAPI needs to handle both v1 and v2 models in the same application

[chbndrhnns]

@YuriiMotov How do you suggest to proceed here?

[github-actions]

This pull request has a merge conflict that needs to be resolved.

[chbndrhnns]

@YuriiMotov I am willing to rebase if we have a clear path to merging

[YuriiMotov]

@YuriiMotov I am willing to rebase if we have a clear path to merging

I think there is no need for this now.
Let's wait for Sebastian to review what we have here and decide if it's the right way to go

[tiangolo]

Thanks for the interest and effort here! This was covered by #14168, which was just released in FastAPI 0.119.0. 🎉

That adds support for pydantic.v1 models at the same time as v2 models. The new docs are here: https://fastapi.tiangolo.com/how-to/migrate-from-pydantic-v1-to-pydantic-v2/

Given that, I'll close this one now. Thanks for the effort and interest and thanks all for the discussions! 🍰