355. Design Twitter §

Date: 2026-02-02 Difficulty: Medium Topics: Heap, Design, Hash Map Link: https://leetcode.com/problems/design-twitter/

⚠️ TODO: Practice this problem again - had many bugs during implementation.

Final Solution §

class Twitter:
 
    def __init__(self):
        self.user_to_following = defaultdict(list)
        self.counter = 0
        self.user_to_tweet = defaultdict(list)
 
        
    def postTweet(self, userId: int, tweetId: int) -> None:
        self.counter += 1
        self.user_to_tweet[userId].append((self.counter, tweetId))
 
    def getNewsFeed(self, userId: int) -> List[int]:
        following = self.user_to_following.get(userId, []) + [userId]
        max_heap = []
        news_feed = []
        for followee in following:
            user_tweets = self.user_to_tweet[followee]
            if user_tweets != []:
                idx = len(user_tweets) - 1
                ts, last_tweet = user_tweets[idx]
                heapq.heappush(max_heap, (-ts, last_tweet, followee, idx))
 
        while max_heap and len(news_feed) < 10:
            ts, last_tweet, followee, index = heapq.heappop(max_heap)
            news_feed.append(last_tweet)
            curr_user_tweets = self.user_to_tweet[followee]
            index -= 1
            if index >= 0:
                next_ts, next_tweet = curr_user_tweets[index]
                heapq.heappush(max_heap, (-next_ts, next_tweet, followee, index))
        
        return news_feed
 
    def follow(self, followerId: int, followeeId: int) -> None:
        if followeeId not in self.user_to_following[followerId]:
            self.user_to_following[followerId].append(followeeId)
 
    def unfollow(self, followerId: int, followeeId: int) -> None:
        if followeeId in self.user_to_following[followerId]:
            self.user_to_following[followerId].remove(followeeId)

Initial Issues (Many!) §

1. heapify returns None: Tried max_heap = heapq.heapify(...)

2. Wrong assignment: self.user_to_tweet = [] instead of self.user_to_tweet[userId] = []

3. Can’t concatenate list + int: following + userId instead of following + [userId]

4. Destroying original data: Used .pop() on tweet list instead of using index

5. Wrong loop condition: while heap or len < 10 instead of while heap and len < 10

6. Inconsistent variable names: min_heap vs max_heap

7. Using old timestamp: Pushed old ts instead of new tweet’s timestamp

8. Wrong indexing: user_tweets[-1][idx] double-indexed instead of user_tweets[idx]

9. Undefined variables: index instead of idx, next_latest_tweet instead of next_tweet

10. Duplicate follows: Could add same followee twice without checking

Key Learnings §

This is Merge K Sorted Lists §

Each user has their own sorted list of tweets (by timestamp). getNewsFeed merges tweets from user + all followees to get the 10 most recent.

Data Structures Needed §

user_to_following: {userId: [followeeIds]}  # Who does this user follow
user_to_tweet: {userId: [(ts, tweetId), ...]}  # User's tweets in chronological order
counter: int  # Global timestamp for ordering tweets

Heap Entry for Merge §

(-timestamp, tweetId, userId, index)
#    ^          ^        ^       ^
#    |          |        |       +-- Where to get next tweet from this user
#    |          |        +---------- Which user this tweet belongs to
#    |          +------------------- The tweet to add to feed
#    +------------------------------ Negated for max heap (most recent first)

Don’t Modify Original Data §

Use an index pointer instead of popping from the original list:

# WRONG - destroys data
user_tweets.pop()
 
# RIGHT - use index
idx = len(user_tweets) - 1
ts, tweet = user_tweets[idx]
idx -= 1  # Move to next tweet

Nuances to Remember §

• Check for duplicate follows before adding
• Check if followee exists before removing (unfollow)
• Use defaultdict to avoid key existence checks
• Include the user themselves in the feed (not just who they follow)
• Store index in heap entry to know where to get next tweet

Q&A Highlights §

Q: Should we precompute feeds or compute on-demand? A: On-demand is better. Precomputing means updating all followers’ feeds on every post, and rebuilding on follow/unfollow.

Q: Why is this merge k sorted lists? A: Each user has a sorted list of tweets. getNewsFeed merges lists from user + followees, taking the 10 most recent overall.