Pydantic

Pydantic is a Python library for data validation and settings management that leverages Python type annotations. It provides a way to declare data models with typed fields and then validate, coerce, and parse raw data into those models, producing structured error information when validation fails.

At the core is the BaseModel class. Fields are defined with type hints, and when data is passed to a model, Pydantic validates the input, converts values to the declared types, applies defaults, and constructs a model instance. If validation cannot be performed, a ValidationError is raised detailing the fields that failed and why.

Validation and customization are central features. Pydantic supports validators via the @validator decorator, allowing pre- and post-processing logic. It handles nested models, lists, dictionaries, unions, and optional fields, and offers features such as field aliases, default factories, and configurable handling of extra fields. Serialization is also supported, with methods to convert models back to plain data structures or JSON. In Pydantic v1, these are typically dict() and json(); in v2, model_dump() and model_dump_json() are provided, with options to include aliases or exclude certain fields.

Usage and ecosystem: Pydantic is widely used for API data validation and configuration management. It is the foundation for request and response models in several web frameworks and is known for its clear error reporting and strong typing discipline. The project is open-source and MIT-licensed, maintained by a community of contributors.

History: Pydantic was created by Samuel Colvin and published to simplify runtime data validation in Python. The library has evolved through multiple releases, with a major rewrite in Pydantic v2 around 2023, focusing on performance improvements and API refinements while keeping backward-compatible concepts.