Sum of Good Subarrays (Adjacent Diff ±1)

MLE L5 coding round: sum over all contiguous subarrays whose elements form a strictly monotonic ±1 step sequence (every adjacent pair differs by exactly +1, OR every adjacent pair differs by exactly -1). Required time complexity is O(N). Requirements Input: integer array arr of length N. A subarray is good if its consecutive differences are either all +1 (monotone increasing by 1) or all -1 (monotone decreasing by 1). Singleton subarrays are always good. Output: the sum of element-sums across all good subarrays (each good subarray contributes the sum of its elements once). Required: O(N) time. Examples arr = [3, 5, 6, 7, 6] Good subarrays: singletons: [3], [5], [6], [7], [6] monotone +1 runs: [5, 6], [6, 7], [5, 6, 7] monotone -1 runs: [7, 6] Not good: [3, 5] (diff 2), [6, 7, 6] (mixes +1 then -1). Notes Walk the array maintaining two run lengths: upRun (current trailing run of +1 steps) and downRun (current trailing run of -1 steps). At index i, only one of them is nonzero (the other resets to 1). Each index i ends upRun + downRun - 1 good subarrays (subtracting the double-counted singleton). For each ending position, contribute the running prefix-sum tail of the active run. Multiple candidates report getting stuck — verbalize the "contribution per index" framing out loud before coding. Easy to mis-handle the reset boundary; double-check with a 2-element example before writing the loop. Preparation Practice the "contribution-per-index" pattern: the sum-of-subarray-minimums and sum-of-subsequence-widths problems both use the same technique. Practice splitting an array into maximal monotone-step runs, then computing per-run contributions in O(1) amortized using prefix sums of element values. Sanity-check with arr = [1, 2, 3] and arr = [3, 2, 1] by hand — both should sum to 20.