dm

from corio import dm

dm provides the data modelling layer built on Pydantic, with utilities aimed at robust parsing and schema ergonomics.

Highlights:

• Field: reusable field definitions with optional auto-defaulting and title/description templating
• Base: model base that can aggregate FIELDS declarations across inheritance
• MixinFromJson: tolerant from_json flow (especially useful for LLM output cleanup)
• to_df: quick conversion to tabular representation (tabular extra needed there)

Install:

pip install "corio[dm]" --upgrade

Typical Model Pattern

from corio import dm


class User(dm.Base):
    id: int
    name: str


user = User.from_json('{"id": 1, "name": "Ada"}')

Field Class Pattern

dm.Field is useful when you want reusable field components across multiple models:

from corio import dm


class FieldName(dm.Field):
    ANNOTATION = str
    DESCRIPTION = "Human-readable user name."


class User(dm.Base):
    FIELDS = [FieldName]
    id: int

LLM/Unclean JSON Use

MixinFromJson routes through json_fix when available, so model parsing can be more tolerant than strict model_validate_json.

This is especially useful for:

• tool-call payload recovery
• partially malformed JSON responses
• low-friction CLI/data ingestion flows

CLI Integration

CliRunMixin enables subcommand-style run() flow for settings/models that integrate with pydantic-settings subcommands.