Implement a Heap Priority Queue

Implement a min-priority queue of integers without using a built-in priority-queue or heap data structure. Process operations from left to right:
PUSH inserts values[i].
PEEK returns the current minimum without removing it.
POP removes and returns the current minimum.
Collect and return every value produced by PEEK or POP, in operation order. Every observation occurs while the queue is nonempty. Equal values are allowed; values[i] is ignored for observation operations.

Function
runPriorityQueue(operations: String[], values: int[]) → int[]

Examples
Example 1
operations = ["PUSH","PUSH","PEEK","POP","PEEK"]
values = [5,2,0,0,0]
return = [2,2,5]
The minimum after both insertions is 2. POP returns and removes 2, leaving 5.
Example 2
operations = ["PUSH","PUSH","PUSH","POP","POP","POP"]
values = [3,3,1,0,0,0]
return = [1,3,3]
Repeated values remain separate queue entries and are removed in nondecreasing priority order.
Example 3
operations = ["PUSH","PEEK","PUSH","PEEK"]
values = [-4,0,-7,0]
return = [-4,-7]
The second insertion becomes the new minimum.

Constraints
1 <= operations.length == values.length <= 200000.
Each operation is PUSH, PEEK, or POP.
Every PEEK and POP occurs when the queue is nonempty.
Each inserted value is a signed 32-bit integer.

Build an array-backed binary min-heap. PUSH appends the value at the end and sifts up while it's smaller than its parent. PEEK returns index 0. POP saves index 0, moves the last element to the root, shrinks the array, and sifts down by swapping with the smaller child until the heap property holds. Every operation is O(log n), so 200000 operations is trivial. The common pitfalls are off-by-one child indexes (2i+1 and 2i+2), forgetting to handle a heap of size one on POP, and comparing with subtraction, which overflows on signed 32-bit values. Use direct less-than comparisons. Duplicates need no special handling, since equal values just stay as separate entries. Ignore values[i] for PEEK and POP. If you freeze on the sift-down loop during the live OA, StealthCoder can give you the working version while you stay in control.

If this hits your live OA and you blank, StealthCoder solves it in seconds, invisible to the proctor.