fms_find_unique

[uramirez8707]

Description
Adds a function to string_utils_mod to determine the number of unique strings in a sorted array and adds a test for it. This will be used in the diag_manager update to determine the number of unique diag_fields in a diag_yaml.

How Has This Been Tested?
CI, including the added unit test.

Checklist:

• My code follows the style guidelines of this project
• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• Any dependent changes have been merged and published in downstream modules
• New check tests, if applicable, are included
• make distcheck passes

[bensonr]

Unless I'm misreading the logic, this only works for a sorted array. To do this without a pre-sort, you'd need a nested loop structure. You can test this by moving the fms_find_unique in your unit test to be line 59 prior to the sort. If you want to stick with a single loop, you'd need add an fms_sort_this prior to the search-loop.

[uramirez8707]

This was only intended to work with a sorted array:

FMS/string_utils/fms_string_utils.F90

Lines 71 to 81 in 3db45b1

	!> @brief c function that finds the number of unique strings in a SORTED array of c pointers
	!! @return number of unique strings
	function fms_find_unique(my_pointer, p_size) bind(c)&
	result(ntimes)
	use iso_c_binding
	type(c_ptr), intent(in) :: my_pointer(*) !< Array of sorted c pointer
	integer(kind=c_int), intent(in) :: p_size !< Size of the array
	integer(kind=c_int) :: ntimes
	end function fms_find_unique

But the documentation should be more clear.

[bensonr]

You could add a call to the fms_sort_this function to ensure a sort is done prior to the comparison search. That would remove the requirement to pass in only sorted arrays.

[uramirez8707]

I added the call

[thomas-robinson]

Is there a quick way to test if it's sorted? Is the sorting of a sorted array fast enough that it doesn't matter that you sort it again?

[thomas-robinson]

The unused printfs should be cleaned up.

[uramirez8707]

Fixed

[thomas-robinson]

These C function variables should be documented https://www.doxygen.nl/manual/commands.html#cmdparam
You also need a \return for the return value

[uramirez8707]

Fixed, I think

[ngs333]

@uramirez8707
The fms_sort_this function uses qsort function from stdlib.h. I was able to look at the code and web comments and it seems to me that it uses the known/published tricks to avoid the worst (O(N^2)) cases. But I am not 100% sure. To be closer to sure, one can check if he array is already sorted or reverse is sorted. This is fast (O(N)) in compariosn to qsort. One can also test and determine the performance as functin of size of input, with test data in sorted order, reverse sorted order, and random order.