Пример #1
0
def hash_columns(columns, result):
    """Hash the *columns* and store in *result*.
    Returns *result*
    """
    assert len(columns) > 0
    assert result.dtype == np.int32
    # No-op for 0-sized
    if len(result) == 0:
        return result
    col_input = [col.cffi_view for col in columns]
    col_out = result.cffi_view
    ncols = len(col_input)
    hashfn = libgdf.GDF_HASH_MURMUR3
    libgdf.gdf_hash(ncols, col_input, hashfn, col_out)
    return result
Пример #2
0
def hash_columns(columns, result, initial_hash_values=None):
    """Hash the *columns* and store in *result*.
    Returns *result*
    """
    assert len(columns) > 0
    assert result.dtype == np.int32
    # No-op for 0-sized
    if len(result) == 0:
        return result
    col_input = [col.cffi_view for col in columns]
    col_out = result.cffi_view
    ncols = len(col_input)
    hashfn = libgdf.GDF_HASH_MURMUR3
    if initial_hash_values is None:
        initial_hash_values = ffi.NULL
    else:
        initial_hash_values = unwrap_devary(initial_hash_values)
    libgdf.gdf_hash(ncols, col_input, hashfn, initial_hash_values, col_out)
    return result