This recipe shines when you are being handed an unstructured data stream that follows a certain pattern. For example, sometimes some old APIs that don't use JSON, or other programs you might want to interact with, hand you streams of data that are grouped by position, like the data we stored in grouped_stream[15], which follows the following pattern:

person0 height0 weight0 person1 height1 weight1 person2 height2 weight2

step_by lets us parse this structure very easily. It works by handing you current element, and then skipping a certain amount of elements on every iteration. In our example, we parse grouped_stream by first creating an iterator over every substring that is not whitespace with split_whitespace [17], then, because we are only interested in the weights, skip the first two elements ("Aaron" and "182cm"), which places our iterator at "70kg". We then tell the iterator to only look at every third element from now on with step_by(3) [19], resulting in us iterating over "70kg", "90kg", and "83kg". Finally, we collect the elements into a vector [20].