Code Generation CLI

EzConfy can generate standalone Python files with Pydantic model definitions from your schema. This gives you editor autocompletion, type checking, and documentation for your config structure — without writing the models by hand.

---

Usage

ezconfy generate schema.yaml -o models.py

This reads schema.yaml and writes a Python file with BaseModel classes matching the schema.

---

Options

Flag	Default	Description
schema_path	(required)	Path to the YAML schema file
-o, --output	generated.py	Output file path

---

Example

schema.yamlGenerated output

types:
  OptimizerType:
    - adam
    - sgd

  TrainingConfig:
    lr: float = 0.001
    epochs: int = 10
    optimizer: OptimizerType

schema:
  training: TrainingConfig
  name: str
  debug: bool?

from enum import Enum

from pydantic import BaseModel, ConfigDict


class OptimizerType(Enum):
    V0 = "adam"
    V1 = "sgd"


class TrainingConfig(BaseModel):
    model_config = ConfigDict(arbitrary_types_allowed=True)

    lr: float = 0.001
    epochs: int = 10
    optimizer: OptimizerType


class ConfigModel(BaseModel):
    model_config = ConfigDict(arbitrary_types_allowed=True)

    training: TrainingConfig
    name: str
    debug: bool | None = None

---

Why Generate Code?

• Editor autocompletion

  ---

  Your IDE knows every config field and its type.

• Static analysis

  ---

  mypy and other type checkers can verify your config usage.

• Documentation

  ---

  The generated file acts as a readable reference for your config structure.

• Refactoring safety

  ---

  Renaming a field in the schema and regenerating catches all usages.

---

Workflow

A typical workflow:

1. Define or update schema.yaml
2. Run ezconfy generate schema.yaml -o config_models.py
3. Import the generated models for type hints:

from config_models import ConfigModel
from ezconfy import ConfigBuilder

cfg: ConfigModel = ConfigBuilder.from_files(  # type: ignore[assignment]
    config_paths="config.yaml",
    schema_path="schema.yaml",
)

# Now your editor knows about cfg.training.lr, cfg.name, etc.

Tip

Regenerate whenever the schema changes to keep your type hints in sync.