Yuqora helps correct Russian and English keyboard layout errors, typos and recurring input mistakes. We are developing our own text input component and shared correction core to improve editor integration and prepare a port to Debian.
Projects
Desktop applications · Text processing
Typing with fewer interruptions.
Yuqora helps correct Russian and English keyboard layout errors, typos and recurring input mistakes. We are developing our own text input component and shared correction core to improve editor integration and prepare a port to Debian.
Windows version · native input component in development
inputrulescorrection
01
Challenge
When working in two languages, the wrong keyboard layout turns a familiar word into an unintended sequence of characters, while typos require revisiting text. Recognising the intended word is only part of the task: an assistant must change the right fragment without affecting subsequent typing after the user moves the caret or switches windows.
02
The existing Windows version
The early Windows version combines Hunspell-based layout and typo correction, personal replacement rules, configurable shortcuts and undo for the last automatic correction. Its interface supports testing rules and inspecting input handler status; separate command groups can launch applications, scripts and HTTP requests. The application uses Electron, with AutoHotkey v2 handling input. Dictionary-based correction runs locally. A test build can be requested through the product website.
03
Our own input component
We are building our own input component to replace the dependency on AutoHotkey. On Windows, we are testing a native C++ component using Text Services Framework, the operating system mechanism for working with text fields. It passes the observed fragment to the shared engine and rechecks the text, caret and selection before applying a correction. The aim is to apply a change only to the editor state it was calculated for, then confirm the actual result.
04
A shared core and the path to Debian
We separate correction decisions from input observation and text editing in a particular operating system. Dictionaries, rules and shared checks belong to the core; a platform adapter handles interaction with editors. This provides a foundation for a Debian version. Linux needs its own adapter and testing in its graphical environments; the Windows TSF component cannot simply be transferred. A Debian version has not yet been released.
05
What has been verified
The new path from editor observation through the shared engine to a text replacement has been tested in selected real-typing scenarios in an isolated Windows environment. Application integration, selection handling, undo, shortcuts and correction triggers remain in development. Broader editor compatibility and an installer containing the new input component still require verification. The existing Windows version and this development are at different stages of readiness.
06
Further development
Further work includes a Debian port, the first free release and inclusion in the Russian Software Register. We are also researching local neural models for text correction and comparing them with the dictionary algorithm on an error corpus; these models are not part of the published early version. Release and registration dates have not been set.
From company records
Materials
Project in the timeline
Start with your challenge