$ show ostrosco/device_query
device_query is a Rust library for querying keyboard and mouse state without requiring an active window, supporting Windows, Linux (X11), and macOS. It provides two primary interfaces: polling-based access via DeviceState (get_mouse() and get_keys()) and event-based callbacks via DeviceEventsHandler for real-time input handling. The architecture features platform-specific implementations abstracted behind a common DeviceState trait, with an event system using background threads to poll for changes and invoke user-provided callbacks with automatic deregistration via CallbackGuard. External dependencies include platform-specific crates (x11 for Linux, windows for Windows, readkey/readmouse/macos-accessibility-client for macOS) and build-time pkg-config for X11 linkage on Linux. The library demonstrates solid cross-platform design and idiomatic Rust patterns. The event system's callback management stands out as a reusable component, using weak references and scoped guards to prevent memory leaks and automatically clean up expired callbacks. This pattern is implemented in both keyboard and mouse callback modules, allowing safe callback registration without manual deregistration. While the core functionality is straightforward, the callback system exhibits non-obvious skill in lifetime management and thread safety. The library is actively maintained with recent commits and issue engagement, though dependencies show mild aging (e.g., windows crate at v0.48.0) that may require updates for newer SDK compatibility but currently function correctly due to stable API usage.
The LLM identified 1 files worth highlighting for reuse:
Manages keyboard callbacks using weak references to prevent memory leaks and automatically removes expired callbacks. Demonstrates a reusable pattern for callback management with automatic cleanup via weak references and scoped guards (CallbackGuard). The same pattern is used for mouse callbacks in mouse_callback.rs, making this a generic solution for safe callback registration in multi-threaded environments.