How do I practice coding and algorithm questions?

Use PracHub's coding console to write, test, and debug your solutions in Python or JavaScript. View hints, test against sample inputs, and compare with official solutions.

What difficulty level is this coding question?

This is a easy difficulty Coding & Algorithms question, commonly asked during Technical Screen rounds at Atlassian.

What role is this question designed for?

This question is commonly asked for Software Engineer candidates at Atlassian during technical interviews.

Filter Invalid Data Events | Atlassian Coding Question

Quick Overview

This question evaluates a candidate's ability to implement data validation, deduplication, stable filtering, time-range and allow-list checks, and to reason about time and space complexity for both batch and streaming scenarios.

Filter Invalid Data Events

Company: Atlassian

Role: Software Engineer

Category: Coding & Algorithms

Difficulty: easy

Interview Round: Technical Screen

You are given a list of data events. Each event contains: - `event_id`: string - `creator_id`: string - `event_type`: string - `event_time`: Unix timestamp - `payload`: arbitrary object Implement a function that returns only the valid events while preserving the original order. An event is considered invalid and must be filtered out if any of the following is true: 1. `event_id` is missing or empty. 2. `creator_id` is missing or empty. 3. `event_time` is missing or malformed. 4. `event_time` is outside a provided inclusive range `[start_time, end_time]`. 5. `event_type` is not in a provided allow-list. 6. The same `event_id` has already appeared earlier; keep only the first valid occurrence. Return the filtered list of events. Follow-up: What are the time and space complexities, and how would your approach change if the events arrived continuously as a stream instead of as a batch?

Quick Answer: This question evaluates a candidate's ability to implement data validation, deduplication, stable filtering, time-range and allow-list checks, and to reason about time and space complexity for both batch and streaming scenarios.

Part 1: Filter Invalid Data Events from a Batch

You are given a batch of data events and must return only the valid ones while preserving their original order. Each event is a dictionary that may contain the keys 'event_id', 'creator_id', 'event_type', 'event_time', and 'payload'. An event is invalid if: (1) 'event_id' is missing or empty, (2) 'creator_id' is missing or empty, (3) 'event_time' is missing or malformed, (4) 'event_time' is outside the inclusive range [start_time, end_time], (5) 'event_type' is not in the allow-list, or (6) the same 'event_id' has already been accepted earlier in the batch. Keep only the first valid occurrence of each event_id. If an earlier event with the same event_id is invalid, it does not block a later valid one.

Constraints

0 <= len(events) <= 200000
0 <= len(allowed_event_types) <= 10000
start_time <= end_time
An event_time is considered valid only if it is an integer Unix timestamp; booleans should be treated as malformed
Missing keys count as missing fields

Examples

Input: ([{'event_id': 'e1', 'creator_id': 'u1', 'event_type': 'click', 'event_time': 10, 'payload': {'x': 1}}, {'event_id': '', 'creator_id': 'u2', 'event_type': 'click', 'event_time': 11, 'payload': None}, {'event_id': 'e2', 'creator_id': 'u3', 'event_type': 'purchase', 'event_time': 15, 'payload': {}}, {'event_id': 'e1', 'creator_id': 'u4', 'event_type': 'click', 'event_time': 12, 'payload': 'dup'}, {'event_id': 'e3', 'creator_id': 'u5', 'event_type': 'view', 'event_time': 21, 'payload': 0}, {'event_id': 'e4', 'creator_id': 'u6', 'event_type': 'view', 'event_time': 20, 'payload': []}], ['click', 'view'], 10, 20)

Loading coding console...

Input: ([{'event_id': 'a', 'creator_id': 'u', 'event_type': 'view', 'event_time': 5, 'payload': None}, {'event_id': 'b', 'creator_id': 'u', 'event_type': 'view', 'event_time': True, 'payload': None}, {'event_id': 'c', 'creator_id': None, 'event_type': 'view', 'event_time': 10, 'payload': None}, {'event_id': 'd', 'creator_id': 'u', 'event_type': 'view', 'event_time': 15, 'payload': None}], ['view'], 5, 15)

Expected Output: [{'event_id': 'a', 'creator_id': 'u', 'event_type': 'view', 'event_time': 5, 'payload': None}, {'event_id': 'd', 'creator_id': 'u', 'event_type': 'view', 'event_time': 15, 'payload': None}]

Explanation: Boundary timestamps 5 and 15 are valid. True is treated as malformed, and creator_id=None is invalid.

Part 2: Streaming Event Acceptance Decisions

In a streaming system, events arrive one at a time and you must decide immediately whether to accept or reject each event. Simulate that online behavior. Given a list of events in arrival order, return a list of booleans where the i-th value is True if the i-th event should be accepted when it arrives, and False otherwise. Use the same rules as the batch version: reject events with missing or empty 'event_id' or 'creator_id', missing or malformed 'event_time', out-of-range timestamps, disallowed 'event_type', or duplicate 'event_id' values that were already accepted earlier. A previous invalid event with the same event_id does not make later valid events duplicates. This models the state you would keep in a real stream processor.

Constraints

0 <= len(events) <= 200000
0 <= len(allowed_event_types) <= 10000
start_time <= end_time
An event_time is considered valid only if it is an integer Unix timestamp; booleans should be treated as malformed
For exact duplicate detection in an unbounded real stream, the set of accepted event_ids can grow without bound

Examples

Input: ([{'event_id': 'a', 'creator_id': 'u1', 'event_type': 'click', 'event_time': 5, 'payload': None}, {'event_id': 'b', 'creator_id': '', 'event_type': 'click', 'event_time': 6, 'payload': None}, {'event_id': 'a', 'creator_id': 'u2', 'event_type': 'click', 'event_time': 7, 'payload': None}, {'event_id': 'c', 'creator_id': 'u3', 'event_type': 'view', 'event_time': 11, 'payload': None}, {'event_id': 'd', 'creator_id': 'u4', 'event_type': 'view', 'event_time': 10, 'payload': None}], ['click', 'view'], 5, 10)

Expected Output: [True, False, False, False, True]

Explanation: The first and last events are accepted. The others fail due to empty creator_id, duplicate accepted event_id, and out-of-range timestamp.

Input: ([{'event_id': 'z1', 'creator_id': 'u1', 'event_type': 'click', 'event_time': None, 'payload': None}, {'event_id': 'z1', 'creator_id': 'u1', 'event_type': 'click', 'event_time': 100, 'payload': {'ok': True}}], ['click'], 0, 200)

Expected Output: [False, True]

Explanation: An invalid event does not reserve its event_id, so the later valid event is accepted.

Input: ([], ['view'], 0, 10)

Expected Output: []

Explanation: Edge case: no arrivals means no decisions.

Input: ([{'event_id': 'm1', 'creator_id': 'u', 'event_type': 'view', 'event_time': 1, 'payload': None}, {'event_id': 'm2', 'creator_id': 'u', 'event_type': 'view', 'event_time': True, 'payload': None}, {'event_id': 'm3', 'creator_id': None, 'event_type': 'view', 'event_time': 2, 'payload': None}, {'event_id': 'm4', 'creator_id': 'u', 'event_type': 'view', 'event_time': 3, 'payload': None}], ['view'], 1, 3)

Expected Output: [True, False, False, True]

Explanation: Boundary timestamps are accepted, but True is malformed as a timestamp and creator_id=None is invalid.

Hints

Think about what state must survive between arrivals. You do not need to store all past events.
A hash set of accepted event_ids is enough for exact duplicate checks, but only add an ID after accepting the event.