langchain-ai/langchain · error · ValueError

search_type of {search_type} not allowed. Expected search_ty

Error message

search_type of {search_type} not allowed. Expected search_type to be 'similarity', 'similarity_score_threshold' or 'mmr'.

What it means

`VectorStore.search(query, search_type, ...)` dispatches on the `search_type` string; anything other than `'similarity'`, `'similarity_score_threshold'`, or `'mmr'` raises `ValueError`. The strictness exists because a typo would otherwise silently fall through to no search at all.

Source

Thrown at libs/core/langchain_core/vectorstores/base.py:324

        Raises:
            ValueError: If `search_type` is not one of `'similarity'`,
                `'mmr'`, or `'similarity_score_threshold'`.
        """
        if search_type == "similarity":
            return self.similarity_search(query, **kwargs)
        if search_type == "similarity_score_threshold":
            docs_and_similarities = self.similarity_search_with_relevance_scores(
                query, **kwargs
            )
            return [doc for doc, _ in docs_and_similarities]
        if search_type == "mmr":
            return self.max_marginal_relevance_search(query, **kwargs)
        msg = (
            f"search_type of {search_type} not allowed. Expected "
            "search_type to be 'similarity', 'similarity_score_threshold'"
            " or 'mmr'."
        )
        raise ValueError(msg)

    async def asearch(
        self, query: str, search_type: str, **kwargs: Any
    ) -> list[Document]:
        """Async return docs most similar to query using a specified search type.

        Args:
            query: Input text.
            search_type: Type of search to perform.

                Can be `'similarity'`, `'mmr'`, or `'similarity_score_threshold'`.
            **kwargs: Arguments to pass to the search method.

        Returns:
            List of `Document` objects most similar to the query.

        Raises:
            ValueError: If `search_type` is not one of `'similarity'`,

View on GitHub (pinned to e32fa9a52e)

Solutions

Set `search_type` to exactly one of `'similarity'`, `'similarity_score_threshold'`, or `'mmr'`.
Validate/normalize external input before passing it: lowercase it and check membership in the allowed set.
For custom search modes, call the underlying method directly (e.g. `store.max_marginal_relevance_search(...)`) instead of the dispatcher.

Example fix

# before
docs = store.search(query, search_type="similarity_score")  # ValueError

# after
docs = store.search(query, search_type="similarity_score_threshold", score_threshold=0.5)

Defensive patterns

Strategy: validation

Validate before calling

ALLOWED = {"similarity", "similarity_score_threshold", "mmr"}

search_type = (search_type or "similarity").lower()
if search_type not in ALLOWED:
    raise ValueError(f"search_type must be one of {sorted(ALLOWED)}, got {search_type!r}")
docs = store.search(query, search_type, **search_kwargs)

Type guard

def is_valid_search_type(value: str) -> bool:
    """Type/narrowing guard for VectorStore.search dispatch values."""
    return isinstance(value, str) and value in {
        "similarity", "similarity_score_threshold", "mmr"
    }

Try / catch

try:
    docs = store.search(query, search_type)
except ValueError as e:
    if "search_type" in str(e):
        docs = store.search(query, "similarity")  # safe fallback
    else:
        raise

Prevention

Normalize external `search_type` input (strip, lowercase) and whitelist-check it.
Centralize the allowed-values constant next to your config parsing so UI and config share one source of truth.
For custom modes, call the specific search method directly instead of the dispatcher.

When it happens

Trigger: Calling `store.search(q, search_type='similiarity')` (typo), `search_type='similarity_score'`, or a custom type the base class does not know; passing user-supplied config directly as `search_type`.

Common situations: Config files with misspelled search types; version drift where an integration once accepted an extra type; UI dropdowns that emit values not matching the three allowed strings; case sensitivity issues (`'MMR'`).

Related errors

AI-assisted analysis of langchain-ai/langchain@e32fa9a52e (2026-08-14). Data as JSON: /api/errors/8ad23f9f67b9c5a9. Report an issue: GitHub.