(2.4.0) JSON Schema for Generics models with Literal as agrument is broken

[bogdandm]

Initial Checks

• I confirm that I'm using Pydantic V2

Description

Straightforward example:

pydantic v2.3.0

{'$defs': {'G_Literal_str__str__': {'properties': {'foo': {'enum': ['a', 'b'],
                                                           'title': 'Foo',
                                                           'type': 'string'}},
                                    'required': ['foo'],
                                    'title': 'G[Literal[str, str]]',
                                    'type': 'object'},
           'M': {'properties': {'foo': {'enum': ['a', 'b'],
                                        'title': 'Foo',
                                        'type': 'string'}},
                 'required': ['foo'],
                 'title': 'M',
                 'type': 'object'}}}

pydantic v2.4.0

{'$defs': {'G': {'properties': {'foo': {'title': 'Foo'}},
                 'required': ['foo'],
                 'title': 'G',
                 'type': 'object'},
           'M': {'properties': {'foo': {'enum': ['a', 'b'],
                                        'title': 'Foo',
                                        'type': 'string'}},
                 'required': ['foo'],
                 'title': 'M',
                 'type': 'object'}}}

Example Code

from pprint import pprint
from pydantic import BaseModel, RootModel
from pydantic.json_schema import models_json_schema
from typing import Generic, TypeVar, Literal

T = TypeVar('T')

class G(BaseModel, Generic[T]):
	foo: T

class M(BaseModel):
	foo: Literal['a', 'b']

pprint(models_json_schema([
	(G[Literal['a', 'b']], 'serialization'),
	(G[Literal['a', 'b']], 'validation'),
	(M, 'validation'),
])[1])

Python, Pydantic & OS Version

venv ❯ python -c "import pydantic.version; print(pydantic.version.version_info())"
             pydantic version: 2.4.0
        pydantic-core version: 2.10.0
          pydantic-core build: profile=release pgo=true
                 install path: /opt/projects/tradingview/venv/lib/python3.11/site-packages/pydantic
               python version: 3.11.5 (main, Aug 25 2023, 13:19:50) [GCC 11.4.0]
                     platform: Linux-6.2.0-33-generic-x86_64-with-glibc2.35
             related packages: typing_extensions-4.7.1 mypy-1.0.1

Works if I downgrade to
             pydantic version: 2.3.0
        pydantic-core version: 2.6.3

[bogdandm]

Actually I think all generics lost its type arguments:

pprint(models_json_schema([
	(G[int], 'serialization'),
	(G[float], 'validation'),
	(M, 'validation'),
])[1])

does not generate any type for field foo too

[hramezani]

Thanks @bogdandm for reporting this 🙏

We will provide a fix for this

[hramezani]

@bogdandm Could you please test this PR to see whether it fixes your problem.

[bogdandm]

Thank you, it is definetly working now. Also it partially fixes #7376 - no more G_Literal_str__str__ in schema 🥳

{'$defs': {'G_Literal__a____b___': {'properties': {'foo': {'enum': ['a', 'b'],
                                                           'title': 'Foo',
                                                           'type': 'string'}},
                                    'required': ['foo'],
                                    'title': "G[Literal['a', 'b']]",
                                    'type': 'object'},
           'M': {'properties': {'foo': {'enum': ['a', 'b'],
                                        'title': 'Foo',
                                        'type': 'string'}},
                 'required': ['foo'],
                 'title': 'M',
                 'type': 'object'}}}

[hramezani]

Thanks @bogdandm for your feedback. it will be available in next release soon