cuhibot

Changelog

This is the history of every major fix and feature added to Cuhi Bot. We try to keep things clear and readable.

Queue Runtime Crash: Fixed NameError crash by properly defining MINIAPP_QUEUE.
Initialization Fix: Correctly passed post_init functions during the bot builder sequence to prevent read-only property assignment errors and ensure the queue worker starts.
Escape Character Rendering: Fixed Markdown formatting where literal backslashes appeared in the UI menu.
Async Safety: Replaced blocking I/O calls (_read_profiles_sync, _read_settings_sync) within the download routines with proper executor-backed awaitable counterparts.
Exception Handling Safety: Removed bare except: blocks that suppressed critical system signals (like KeyboardInterrupt) ensuring graceful shutdowns.
Missing Feedback: Added a proper error message when a user tries to export an empty list of sources.
Server Safety: Replaced unsafe os.environ index with a .get() fallback in server.py to prevent crashes when BOT_TOKEN is missing at load time.

Full Async I/O: Migrated all blocking disk operations (JSON reading/writing, directory scans) to a dedicated thread pool (ThreadPoolExecutor). This makes the bot 100% responsive even during massive multi-user downloads.
Resource Leak Fix: Implemented contextlib.ExitStack in the media group sender to guarantee that every file descriptor is closed properly, preventing “Too many open files” crashes.
Throttled Status: Refined the status update engine to prevent Telegram rate limits and ensure smooth UI transitions.
Atomic Locking: Optimized file locking with monotonic timers for safer concurrent state management.

Native Video Grouping: Videos are now properly packaged into 10-item media groups alongside photos, significantly reducing chat clutter.
Interleaved Downloads: The “Both” download option now intelligently mixes photos and videos together instead of performing two separate passes.
Anti-Rate Limit: Increased safety delays and implemented --continue flags to prevent gallery-dl from dropping out mid-profile due to platform restrictions.
Static Analysis (Pass 18): Performed a deep-dive PEP-8 formatting pass using flake8 to resolve all lingering indentation inconsistencies and ensure standard alignment.

We did a massive deep-dive (Pass 17) to make the bot truly production-ready.

Atomic Profiles: No more data loss. We added OS-level file locking so profile changes are always safe.
Smart Uploads: The bot now pre-checks file sizes. If a file is over 50MB (Telegram’s limit), it skips it and tells you why instead of crashing the whole batch.
Bypass Archive: You can now use /link <url> to re-download things even if they were already archived.
Real-time Fix: The download engine is now much more responsive and handles process cleanup properly on Windows.

This was our 16th audit pass. We focused on cleaning up memory leaks and fixing minor UI glitches in the callback handlers.

We added 5 big features to make the bot more autonomous:

For the full history of earlier versions, see the Git commit logs.

This site is open source. Improve this page.