Optimisation of n square array node comparison in LenientJsonArrayPartialMatcher

[aymar99]

LenientJsonArrayPartialMatcher performs a comparison of each element in the expected array node with each element in the actual array node, resulting in n^2 complexity for calculating the similarity score before identifying the best matching pairs for comparison. Here is the code link.

I identified an opportunity for optimization in this process. By filtering out identical elements code link from both the expected and actual arrays before applying the n^2 similarity score calculation, we can significantly reduce the complexity. In scenarios where there are only a few mismatches between the expected and actual arrays, this optimization ensures that the n^2 complexity is only applied to those differing elements. The worst-case scenario of n^2 for all elements in the array occurs only if none of the elements match.

For smaller JSONs, the implementation may not exhibit a noticeable difference. However, during testing with larger JSONs containing hundreds of array elements, a significant performance improvement becomes apparent. I have tested this enhancement and created a pull request. I would appreciate it if you could review the pull request and share your thoughts!

[deblockt]

Hi @aymar99 , Thanks for this PR.

Sorry for my delayed review, I have see a litle issue on method getElementsWithCount it seem that your never add items on nodeCounter map.
I have rework some code to have lower complexity on the PR #19.

It would be great if you can review it.

Thanks for your help.

[aymar99]

@deblockt the intention of that method was to get elements by count. I remember making it work as a whole. Some mess up has happened and I have pushed the wrong implementation as you pointed out. Saw your pull request and it has correct implementation of the idea proposed in this PR. Thanks for accommodating the proposal.