Function load_provider_config

paperflow/providers/config.py:99–142

Read provider settings from the environment. The function is pure with respect to the current environment — it is fine to call repeatedly, and tests can monkeypatch ``os.environ``.

()

def load_provider_config() -> ProviderConfig:
    """Read provider settings from the environment.

    The function is pure with respect to the current environment — it is
    fine to call repeatedly, and tests can monkeypatch ``os.environ``.
    """

    llm_provider = _normalize_llm_provider(
        _first_env("PAPERFLOW_LLM_PROVIDER", "LLM_PARSER_PROVIDER")
    )
    llm_model = (
        _first_env(
            "PAPERFLOW_LLM_MODEL",
            "LLM_PARSER_OPENAI_MODEL",
            "DASHSCOPE_LLM_MODEL",
            "HF_LLM_MODEL",
        )
        or _DEFAULT_LLM_MODEL[llm_provider]
    )

    embed_provider = _normalize_embed_provider(
        _first_env("PAPERFLOW_EMBED_PROVIDER", "EMBEDDING_PROVIDER")
    )
    embed_model = (
        _first_env("PAPERFLOW_EMBED_MODEL", "EMBEDDING_MODEL")
        or _DEFAULT_EMBED_MODEL[embed_provider]
    )

    raw_dim = os.environ.get("PAPERFLOW_EMBED_DIMENSIONS", "").strip()
    if raw_dim:
        try:
            embed_dimensions = int(raw_dim)
        except ValueError:
            embed_dimensions = _DEFAULT_EMBED_DIM[embed_provider]
    else:
        embed_dimensions = _DEFAULT_EMBED_DIM[embed_provider]

    return ProviderConfig(
        llm_provider=llm_provider,
        llm_model=llm_model,
        embed_provider=embed_provider,
        embed_model=embed_model,
        embed_dimensions=embed_dimensions,
    )