daft.functions.minhash#

def minhash(
    text: Expression,
    *,
    num_hashes: int,
    ngram_size: int,
    seed: int = 1,
    hash_function: Literal["murmurhash3", "xxhash", "xxhash32", "xxhash64", "xxhash3_64", "sha1"] = "murmurhash3",
) -> Expression:
    """Runs the MinHash algorithm on the series.

    For a string, calculates the minimum hash over all its ngrams,
    repeating with `num_hashes` permutations. Returns as a list of 32-bit unsigned integers.

    Tokens for the ngrams are delimited by spaces.
    The strings are not normalized or pre-processed, so it is recommended
    to normalize the strings yourself.

    Args:
        text (String Expression): The expression to hash.
        num_hashes (int): The number of hash permutations to compute.
        ngram_size (int): The number of tokens in each shingle/ngram.
        seed (int, default=1): Seed used for generating permutations and the initial string hashes. Defaults to 1.
        hash_function (str, default="murmurhash3"): Hash function to use for initial string hashing. One of "murmurhash3", "xxhash" (alias for "xxhash3_64"), "xxhash32", "xxhash64", "xxhash3_64", or "sha1". Defaults to "murmurhash3".

    Returns:
        Expression (FixedSizedList[UInt32, num_hashes] Expression):
            expression representing the MinHash values.

    """
    return Expression._call_builtin_scalar_fn(
        "minhash", text, num_hashes=num_hashes, ngram_size=ngram_size, seed=seed, hash_function=hash_function
    )