ORDERED_PAIR_INDEXING
 
 Returns the value of the OrderedPair at the requested index.   Params:    default : OrderedPair  The input OrderedPair to index.   index : int  The index of the OrderedPair to return.   x_axis : bool  Index x axis? If not y is indexed.     Returns:    out : Scalar  The scalar index of the input OrderedPair.    
   Python Code
from flojoy import flojoy, OrderedPair, Scalar
@flojoy
def ORDERED_PAIR_INDEXING(
    default: OrderedPair,
    index: int = 0,
    x_axis: bool = False,
) -> Scalar:
    """Returns the value of the OrderedPair at the requested index.
    Parameters
    ----------
    default : OrderedPair
        The input OrderedPair to index.
    index : int
        The index of the OrderedPair to return.
    x_axis : bool
        Index x axis? If not y is indexed.
    Returns
    -------
    Scalar
        The scalar index of the input OrderedPair.
    """
    assert (
        len(default.x) > index
    ), "The index parameter must be less than the length of the OrderedPair."
    assert index >= 0, "The index parameter must be greater than zero."
    if x_axis:
        c = default.x[index]
    else:
        c = default.y[index]
    return Scalar(c=c)
Example
Having problems with this example app? Join our Discord community and we will help you out!
This app uses ordered pair transformation techniques.
First the necessary blocks were added:
- LINSPACE
- SINE
- ORDERED_PAIR_LENGTH
- ORDERED_PAIR_INDEXING
- ORDERED_PAIR_DELETE
- 2x BIG_NUMBER
- TABLE
LINSPACE and SINE created an ordered pair data type (x and y axis pairs). ORDERED_PAIR_LENGTH extracts the length of an ordered pair (this length is viewed with BIG_NUMBER here). ORDERED_PAIR_INDEXING extracts a single value from either the x or y axes (this value is viewed with BIG_NUMBER here). ORDERED_PAIR_DELETE deletes single indexes from both the x or y axes and outputs a new ordered pair (the new ordered pair is viewed with BIG_NUMBER here).
The blocks were connected as shown and the app was run.