Design an API QPS Counter

Implement an in-memory API QPS counter. Record start_time when the server starts. Each request has a current timestamp in seconds (in an actual interview, system time may be used). Support the following operations: put(key, value): records one PUT request. get(key): records one GET request. get_load(): returns the average GET QPS and PUT QPS separately over the last five minutes. QPS Definition Let now be the current timestamp and start_time be the server start timestamp: Count requests in the trailing 300-second window: requests satisfying now - timestamp < 300. The denominator must not always be 300. If the server has been running for fewer than 300 seconds, use its actual running time: elapsed = min(now - start_time, 300) qps = request_count_in_window / elapsed If elapsed = 0, define QPS as 0.0. A get_load() call itself does not count as either a GET or a PUT request. key and value do not affect QPS accounting; you do not need to implement actual key-value storage semantics. Use a fixed-size ring buffer with one bucket per second. Each operation should run in O(1) time and use O(300) space. Input Format For deterministic testing, timestamps are explicitly supplied. The first line contains server start time start_time; the second line contains the number of operations n; each subsequent line is one of: timestamp PUT key value timestamp GET key timestamp LOAD For every LOAD, print: get_qps put_qps Print both values with six decimal places. Example Input: 0 5 1 PUT a 1 2 GET a 3 GET b 10 LOAD 301 LOAD Output: 0.222222 0.111111 0.000000 0.000000 At time 10, the server has run for 10 seconds and the active window contains 2 GETs and 1 PUT. At time 301, all prior requests have expired from the trailing 300-second window. Example Input 0 5 1 PUT a 1 2 GET a 3 GET b 10 LOAD 301 LOAD Output 0.222222 0.111111 0.000000 0.000000