In the random test data, when passing in a 0 value to the battle method, the test is treating that as a valid opponent level score. The instructions state only levels 1-100 are valid; therefore, the expected return value should be "Invalid level".

Additionally on the battle(0) calls, experience points are also being awarded depending the Warrior's current level.

Definitely something wring with the random tests (I failed 20 of them after passing all of the basic tests):

Small Arrays
arr1: [34, 38, 7, 6, 4, 46, 17, 36, 33, 45, 1, 47, 13, 14, 11, 32, 20, 9, 49, 18, 21, 35, 19, 40]
arr2: [33, 2, 30, 46, 15, 5, 20, 11, 0, 24, 50, 6, 37, 16, 13, 29, 44, 32, 21, 1, 12, 35]
✘ Expected: [0, 26, 14, 12], instead got: [10, 26, 14, 12]

The intersection of the two arrays is [6, 46, 33, 1, 13, 11, 32, 20, 21, 35] for a length of 10.

Confirmed with minitest on my local machine:

def test_random_example
arr1 = [34, 38, 7, 6, 4, 46, 17, 36, 33, 45, 1, 47, 13, 14, 11, 32, 20, 9, 49, 18, 21, 35, 19, 40]
arr2 = [33, 2, 30, 46, 15, 5, 20, 11, 0, 24, 50, 6, 37, 16, 13, 29, 44, 32, 21, 1, 12, 35]
assert_equal(process_2arrays(arr1, arr2), [10, 26, 14, 12])
end

Running:

.

Finished in 0.001227s, 814.9959 runs/s, 814.9959 assertions/s.

1 runs, 1 assertions, 0 failures, 0 errors, 0 skips