Memory grows unbounded during long pagination

[sujito00]

Bug

The paginated methods (searchAll, getUserTweetsAll, getBookmarksAll) use a seen Set to deduplicate tweets. This Set accumulates all tweet IDs indefinitely.

File: src/api/client.ts (lines ~514-623)

For getBookmarksAll() with maxPages = Infinity, this could grow to millions of entries on accounts with large bookmark collections.

Expected behavior

Either:

• Use a bounded data structure (e.g., last N IDs) since duplicates typically only appear in adjacent pages
• Stream results instead of accumulating them (yield tweets as they're fetched)
• Add a hard cap with a warning

Impact

Low for typical use. High for power users exporting large datasets (bookmarks --all on accounts with 10k+ bookmarks).