I assume P and S don't know that these four are the only options, otherwise P should know the numbers (the products are 15, 88, 14, 52, all unique).
P can know the numbers if and only if the product P is a prime, otherwise (1,P) and (n,P/n) are options for a non-trivial divisor n. None of the products is a prime, so the first line doesn't help us. Let's look at S' reply:
If the sum is 8 then the product could be 7 (1,7) and P could know the numbers. S knows this cannot be the case, therefore the sum cannot be 8 and the numbers can't be (3,5).
If the sum is 19 then the numbers are (n,19-n) and n(19-n) is not a prime for any n (for n=1 we get 18). Similarly for 9 and 17 as sum. In all three cases S knows P cannot know the numbers. In fact p+1 for some prime p is the only sum where the product might be unique: For all other numbers S knows P cannot know the numbers.
Let's go back to P.
What could the numbers be if the product was 88? (1,88), (2,44), (4,22), (8,11) are all options, leading to sums 89, 46, 26, 19. None of these is one larger than a prime. If the product is 88 then P knows from the beginning on that S knows that they cannot determine the numbers.
What could the numbers be if the product was 14? (1,14) and (2,7) are the only options, leading to sums 15 and 9. Again none of these is one larger than a prime.
What could the numbers be if the product was 52? We have (1,52), (2,26), (4,13) with sums 53,28,17. Same here.
All three options looks possible. I don't see how P could continue to rule out some of them based on what S said.