Traffic-control practice

Opening your lesson

A new service has the same familiar failure modes with different request patterns. Admission, cancellation, retries and acknowledgments each have their own boundary. Choose the contract named in the prompt rather than applying one generic error handler to everything.

This set mixes deterministic reasoning with actual coroutine execution. A sliding-window check differs from the fixed-window project; its left boundary expires old accepted requests. Retry decisions use supplied statuses and never contact a real endpoint. Queue work must preserve data and finish its owned tasks. The examples include exact limits, zero values and empty batches. No new concept is introduced, but the combination requires you to keep several earlier invariants separate while producing one precise result per task.

import json

def process_payload(payload: str) -> str:
    data = json.loads(payload)
    from collections import deque
    accepted = deque()
    result = []
    for now in data['times']:
        while accepted and accepted[0] <= now - data['window']:
            accepted.popleft()
        allowed = len(accepted) < data['limit']
        result.append(allowed)
        if allowed:
            accepted.append(now)
    return json.dumps(result,separators=(',',':'))

Expire accepted arrivals exactly one window old; rejected arrivals do not extend the wait. Store only accepted arrival times. Expire old times before deciding whether the next request fits. Each accepted time enters and leaves the deque once, giving linear total queue work.

Choose your character

Your Work Gym account