Skip to content

INSTALLATION.md

Latest commit

 

History

History
75 lines (55 loc) · 2.62 KB

INSTALLATION.md

File metadata and controls

75 lines (55 loc) · 2.62 KB
Raw

Installation and operation

Download release build

  1. Open the latest release: https://github.com/sbay-dev/MagicMirror/releases/latest
  2. Download MagicMirror-v1.0.5-windows-x64.zip.
  3. Extract the ZIP to a writable folder.
  4. Run MagicMirror.Native.exe.

If Windows SmartScreen appears, choose More info -> Run anyway only if you downloaded the file from the official sbay-dev/MagicMirror release page.

Requirements

  • Windows 10 1809 or newer.
  • .NET 10 runtime/SDK for framework-dependent builds.
  • Windows OCR language packs for native OCR fallback.
  • Optional: local Tesseract installation with eng/ara traineddata. The app auto-discovers tessdata from the Tesseract install, the app folder, or TESSDATA_PREFIX; configure TessDataPath only when auto-discovery fails.

First run

  1. Open the control window.
  2. Choose target language, OCR mode, colors, and optional AI gateway.
  3. Click Open Mirror Overlay.
  4. Move the glass over a document or application.
  5. Click ترجم.
  6. Click a visible word to open/select dictionary context; drag the dictionary card away when it covers the source text.

AI gateway

The native app posts to:

{GatewayBaseUrl}/api/sarmad/ask

When GatewayBaseUrl is empty, unavailable, or returns unusable/wrong-language output, the app keeps the Sarmad/original path unless the user explicitly confirms the MT prompt for that translation run. MT fallback sends text to third-party MT services (Google gtx / MyMemory), is not equivalent to the domain-aware academic Sarmad AI path, and should be declined for confidential manuscripts or publisher-style review. Dictionary alternatives require a configured Sarmad gateway and will report that the gateway is not configured instead of fabricating results.

The old documentation gateway at https://wmr-doc.pages.dev/api/sarmad/ask is not used as a default product fallback because it was deployed against the deprecated @cf/openai/gpt-oss-120b model.

For reliable dictionary and AI translation results, v1.0.4 and newer default to the dedicated Worker:

https://magicmirror-sarmad-gateway.2sa.workers.dev

See CLOUDFLARE_GATEWAY.md if you need to redeploy or replace the gateway.

OCR

The pipeline prefers UI Automation text when the target window exposes it. If no accessible text is available, it uses OCR:

  • Tesseract, when tesseract.exe and tessdata are configured or discoverable.
  • Windows.Media.Ocr as fallback.

Logs

Runtime logs are written to the application data directory as mirror.log. Use the log when reporting capture, OCR, translation, or dictionary failures.