Provides composable, pure functions (pipe, wrap, slugify, romanize, redact, etc.) that transform generated values at call time. Each function is stateless, type-annotated, and importable without any internal Mimesis dependency. The pattern is directly reusable in any data-generation or serialisation pipeline—extract the file and it works immediately.

Implements a custom Random class extending Python's random.Random with additional methods for weighted choices, uniform distribution over integers, and deterministic seeding. The class is entirely self-contained (standard library only) and can be extracted and used in any project requiring reproducible, cryptographically-insecure randomness with custom selection logic.

Defines BaseProvider and BaseDataProvider abstract base classes with an automatic registry pattern (via __init_subclass__ and optional Meta attributes like name, datafile, auto_register). This design allows new providers to be created simply by subclassing and setting Meta; the registry is then discoverable by Generic. The composability and registration mechanics are architecturally clean and could be extracted (with minimal adaptation) to build a plugin-based data-generation framework.