BuilderFactories

class scikitplot.corpus.BuilderFactories(reader_factory=None, chunker_factory=None, filter_factory=None, normalizer_factory=None, enricher_factory=None, embedding_engine_factory=None)

Component factory callables for FactoryCorpusBuilder.

Each factory replaces the corresponding lazy-creation method in CorpusBuilder. None means “use the default from BuilderConfig”.

Parameters:

reader_factorycallable or None, optional

Factory for DocumentReader. Called once per source. Receives (source: str | Path, chunker, **reader_kwargs) -> DocumentReader. Signature:

def reader_factory(
    source: str | pathlib.Path,
    chunker: ChunkerBase | None,
    **reader_kwargs: Any,
) -> DocumentReader: ...

chunker_factorycallable or None, optional

Factory for the chunker. Called once at build time. No arguments. Signature:

def chunker_factory() -> ChunkerBase | None: ...

filter_factorycallable or None, optional

Factory for the FilterBase. Called once at build time. No arguments. Signature:

def filter_factory() -> FilterBase | None: ...

normalizer_factorycallable or None, optional

Factory for the NormalizationPipeline. Called once at build time. No arguments. Signature:

def normalizer_factory() -> NormalizationPipeline | None: ...

enricher_factorycallable or None, optional

Factory for the enricher. Called once at build time. No arguments. Signature:

def enricher_factory() -> NLPEnricher | None: ...

embedding_engine_factorycallable or None, optional

Factory for the embedding engine. Called once at build time. No arguments. Signature:

def embedding_engine_factory() -> EmbeddingEngine | None: ...

Parameters:

• reader_factory (Callable[[...], Any] | None)

• chunker_factory (Callable[[], Any] | None)

• filter_factory (Callable[[], Any] | None)

• normalizer_factory (Callable[[], Any] | None)

• enricher_factory (Callable[[], Any] | None)

• embedding_engine_factory (Callable[[], Any] | None)

Notes

User note: Factories take precedence over the corresponding BuilderConfig settings. For example, if chunker_factory is set, BuilderConfig.chunker is ignored for chunker creation.

Examples

Use a custom reader factory that injects a per-source language code:

from langdetect import detect

def smart_reader_factory(source, chunker, **kw):
    lang = detect(open(source).read(200)) if Path(source).exists() else None
    return DocumentReader.create(source, chunker=chunker, default_language=lang)

factories = BuilderFactories(reader_factory=smart_reader_factory)
builder = FactoryCorpusBuilder(factories=factories)
result = builder.build("./data/")

chunker_factory: Callable[[], Any] | None = None

embedding_engine_factory: Callable[[], Any] | None = None

enricher_factory: Callable[[], Any] | None = None

filter_factory: Callable[[], Any] | None = None

normalizer_factory: Callable[[], Any] | None = None

reader_factory: Callable[[...], Any] | None = None