Discriminator inside list items not snake-cased with --snake-case-field + --collapse-root-models

[TeaDrinkingProgrammer]

Bug report: discriminator inside list items not converted to snake_case

Describe the bug
When using --snake-case-field together with --collapse-root-models, a discriminator
defined on an array's items schema is generated with the original camelCase property
name instead of the converted snake_case field name. This causes Pydantic v2 to raise
a PydanticUserError on import.

This is a partial regression of #1769. The fix there works for discriminators on direct
properties, but not when the discriminator is collapsed from a RootModel wrapping list
items.

To Reproduce

Example schema:

openapi: "3.0.0"
info:
  title: Repro
  version: "1.0"
paths: {}
components:
  schemas:
    FooType:
      type: object
      required: [objectType]
      properties:
        objectType:
          type: string
          enum: [FOO]
    BarType:
      type: object
      required: [objectType]
      properties:
        objectType:
          type: string
          enum: [BAR]
    Container:
      type: object
      properties:
        items:
          type: array
          items:
            discriminator:
              propertyName: objectType
            oneOf:
              - $ref: "#/components/schemas/FooType"
              - $ref: "#/components/schemas/BarType"

Used commandline:

$ datamodel-codegen \
  --input openapi.yaml \
  --input-file-type openapi \
  --output-model-type pydantic_v2.BaseModel \
  --use-annotated \
  --snake-case-field \
  --collapse-root-models \
  --output models.py

Then import the generated file:

$ python -c "import models"

Expected behavior
The generated Container field should use the snake_case field name in the discriminator:

class Container(BaseModel):
    items: (
        list[Annotated[FooType | BarType, Field(discriminator='object_type')]] | None
    ) = None

Importing models.py succeeds.

Actual behavior
The generated discriminator retains the camelCase alias:

class Container(BaseModel):
    items: (
        list[Annotated[FooType | BarType, Field(discriminator='objectType')]] | None
    ) = None

Importing models.py raises:

pydantic.errors.PydanticUserError: Model 'FooType' needs a discriminator field for key 'objectType'

For further information visit https://errors.pydantic.dev/2.12/u/discriminator-no-field

Note: without --collapse-root-models the bug does not occur — the generator wraps
the items in an intermediate RootModel and correctly generates discriminator='object_type' there.

Version:

• OS: Linux
• Python version: 3.14.3
• datamodel-code-generator version: 0.56.1

Additional context
Related to #1769 which fixed discriminators on direct properties. The collapsed
RootModel code path was not covered by that fix.

[coderabbitai]

🔗 Related PRs

#2717 - fix: skip non-model types in __change_field_name [merged]
#2722 - Fix discriminator with allOf without Literal type for Pydantic v2 [merged]
#2853 - Fix reuse-model generating type aliases instead of class inheritance [merged]
#3022 - Extract shared pydantic base module for v2/dataclass/msgspec [merged]
#3070 - Fix non-string OpenAPI discriminator literals [merged]

---

📝 Issue Planner

_{Check the box below or use the @coderabbitai plan command to generate an implementation plan and prompts that you can use with your favorite coding assistant.}

• Create Plan

---

🧪 Issue enrichment is currently in open beta.

You can configure auto-planning by selecting labels in the issue_enrichment configuration.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

💬 Have feedback or questions? Drop into our discord!