Function build_embedding_provider

paperflow/providers/embedding.py:163–198

Construct the configured embedding provider, falling back to hash on missing deps.

(config: Optional[ProviderConfig] = None)

def build_embedding_provider(config: Optional[ProviderConfig] = None) -> EmbeddingProvider:
    """Construct the configured embedding provider, falling back to hash on missing deps."""

    cfg = config or load_provider_config()

    if cfg.embed_provider == "openai":
        api_key = os.environ.get("OPENAI_API_KEY", "")
        if _is_placeholder(api_key):
            return HashEmbedding(dimensions=cfg.embed_dimensions)
        base_url = os.environ.get("OPENAI_BASE_URL") or None
        timeout = float(os.environ.get("OPENAI_API_TIMEOUT", "60") or 60)
        return OpenAIEmbedding(
            model=cfg.embed_model,
            dimensions=cfg.embed_dimensions,
            api_key=api_key,
            base_url=base_url,
            timeout=timeout,
        )

    if cfg.embed_provider == "sentence_transformers":
        try:
            return SentenceTransformersEmbedding(model=cfg.embed_model, dimensions=cfg.embed_dimensions)
        except ImportError:
            return HashEmbedding(dimensions=cfg.embed_dimensions)

    if cfg.embed_provider == "ollama":
        base_url = os.environ.get("OLLAMA_BASE_URL", "http://localhost:11434")
        timeout = float(os.environ.get("OLLAMA_API_TIMEOUT", "60") or 60)
        return OllamaEmbedding(
            model=cfg.embed_model,
            dimensions=cfg.embed_dimensions,
            base_url=base_url,
            timeout=timeout,
        )

    return HashEmbedding(dimensions=cfg.embed_dimensions)


__all__ = [