lib/extractor/ntr_text_fragmentation/chunking/chunking_registry.py

from ..chunking.chunking_strategy import ChunkingStrategy


class _ChunkingRegistry:
    def __init__(self):
        self._chunking_strategies: dict[str, ChunkingStrategy] = {}

    def register(self, name: str, strategy: ChunkingStrategy):
        self._chunking_strategies[name] = strategy

    def get(self, name: str) -> ChunkingStrategy:
        return self._chunking_strategies[name]

    def get_names(self) -> list[str]:
        return list(self._chunking_strategies.keys())

    def __len__(self) -> int:
        return len(self._chunking_strategies)

    def __contains__(self, name: str | ChunkingStrategy) -> bool:
        if isinstance(name, ChunkingStrategy):
            return name in self._chunking_strategies.values()
        return name in self._chunking_strategies

    def __dict__(self) -> dict:
        return self._chunking_strategies

    def __getitem__(self, name: str) -> ChunkingStrategy:
        return self._chunking_strategies[name]


chunking_registry = _ChunkingRegistry()


def register_chunking_strategy(name: str | None = None):
    def decorator(cls: type[ChunkingStrategy]) -> type[ChunkingStrategy]:
        chunking_registry.register(name or cls.__name__, cls)
        return cls

    return decorator