[Extending API] /input

[raiden-staging]

• Extending server with /input routes

# Input features ---
# Mouse

# | POST /input/mouse/move — Move mouse to absolute coordinates
curl -X POST -H "Content-Type: application/json" \
  --data '{"x": 500, "y": 500}' \
  http://localhost:10001/input/mouse/move
# Response: {"ok":true}

# | POST /input/mouse/move_relative — Move mouse relative to current position
curl -X POST -H "Content-Type: application/json" \
  --data '{"dx": 50, "dy": -25}' \
  http://localhost:10001/input/mouse/move_relative
# Response: {"ok":true}

# | POST /input/mouse/click — Click mouse button
curl -X POST -H "Content-Type: application/json" \
  --data '{"button":"left","count":2}' \
  http://localhost:10001/input/mouse/click
# Response: {"ok":true}

# | POST /input/mouse/down — Press mouse button down
curl -X POST -H "Content-Type: application/json" \
  --data '{"button":"left"}' \
  http://localhost:10001/input/mouse/down
# Response: {"ok":true}

# | POST /input/mouse/up — Release mouse button
curl -X POST -H "Content-Type: application/json" \
  --data '{"button":"left"}' \
  http://localhost:10001/input/mouse/up
# Response: {"ok":true}

# | POST /input/mouse/scroll — Scroll mouse wheel
curl -X POST -H "Content-Type: application/json" \
  --data '{"dx":0,"dy":-120}' \
  http://localhost:10001/input/mouse/scroll
# Response: {"ok":true}

# | GET /input/mouse/location — Get current mouse location
curl http://localhost:10001/input/mouse/location
# Response: {"x":500,"y":500,"screen":0,"window":"60817493"}


# Keyboard

# | POST /input/keyboard/type — Type text
curl -X POST -H "Content-Type: application/json" \
  --data '{"text":"Hello, World!","wpm":300,"enter":true}' \
  http://localhost:10001/input/keyboard/type
# Response: {"ok":true}

# | POST /input/keyboard/key — Send key sequence
curl -X POST -H "Content-Type: application/json" \
  --data '{"keys":["ctrl","a"]}' \
  http://localhost:10001/input/keyboard/key
# Response: {"ok":true}

# | POST /input/keyboard/key_down — Press and hold a key
curl -X POST -H "Content-Type: application/json" \
  --data '{"key":"ctrl"}' \
  http://localhost:10001/input/keyboard/key_down
# Response: {"ok":true}

# | POST /input/keyboard/key_up — Release a key
curl -X POST -H "Content-Type: application/json" \
  --data '{"key":"ctrl"}' \
  http://localhost:10001/input/keyboard/key_up
# Response: {"ok":true}


# Window

# | POST /input/window/activate — Activate a window by match
curl -X POST -H "Content-Type: application/json" \
  --data '{"match":{"title_contains":"New Tab","only_visible":true}}' \
  http://localhost:10001/input/window/activate
# Response: {"activated":true,"wid":"60817493"}

# | POST /input/window/focus — Focus a window by match
curl -X POST -H "Content-Type: application/json" \
  --data '{"match":{"class":"Google-chrome"}}' \
  http://localhost:10001/input/window/focus
# Response: {"focused":true,"wid":"60817493"}

# | POST /input/window/move_resize — Move/resize a window by match
curl -X POST -H "Content-Type: application/json" \
  --data '{"match":{"title_contains":"Chrome"},"x":0,"y":0,"width":1280,"height":720}' \
  http://localhost:10001/input/window/move_resize
# Response: {"ok":true}

# | POST /input/window/raise — Raise window to top
curl -X POST -H "Content-Type: application/json" \
  --data '{"match":{"pid":12345}}' \
  http://localhost:10001/input/window/raise
# Response: {"ok":true}

# | POST /input/window/minimize — Minimize window
curl -X POST -H "Content-Type: application/json" \
  --data '{"match":{"title_contains":"New Tab"}}' \
  http://localhost:10001/input/window/minimize
# Response: {"ok":true}

# | POST /input/window/map — Show window
curl -X POST -H "Content-Type: application/json" \
  --data '{"match":{"title_contains":"New Tab"}}' \
  http://localhost:10001/input/window/map
# Response: {"ok":true}

# | POST /input/window/unmap — Hide window
curl -X POST -H "Content-Type: application/json" \
  --data '{"match":{"title_contains":"New Tab"}}' \
  http://localhost:10001/input/window/unmap
# Response: {"ok":true}

# | POST /input/window/close — Close window by match
curl -X POST -H "Content-Type: application/json" \
  --data '{"match":{"title_contains":"New Tab"}}' \
  http://localhost:10001/input/window/close
# Response: {"ok":true,"wid":"60817493","windowIds":["60817493"]}

# | POST /input/window/kill — Force-kill window by match
curl -X POST -H "Content-Type: application/json" \
  --data '{"match":{"title_contains":"Unresponsive App"}}' \
  http://localhost:10001/input/window/kill
# Response: {"ok":true}

# | GET /input/window/active — Get active window
curl http://localhost:10001/input/window/active
# Response: {"wid":"60817493"}

# | GET /input/window/focused — Get focused window
curl http://localhost:10001/input/window/focused
# Response: {"wid":"60817493"}

# | POST /input/window/name — Get window name
curl -X POST -H "Content-Type: application/json" \
  --data '{"wid":"60817493"}' \
  http://localhost:10001/input/window/name
# Response: {"wid":"60817493","name":"New Tab - Google Chrome"}

# | POST /input/window/pid — Get window PID
curl -X POST -H "Content-Type: application/json" \
  --data '{"wid":"60817493"}' \
  http://localhost:10001/input/window/pid
# Response: {"wid":"60817493","pid":42420}

# | POST /input/window/geometry — Get window geometry
curl -X POST -H "Content-Type: application/json" \
  --data '{"wid":"60817493"}' \
  http://localhost:10001/input/window/geometry
# Response: {"wid":"60817493","x":0,"y":0,"width":1280,"height":720,"screen":0}


# Display

# | GET /input/display/geometry — Get display geometry
curl http://localhost:10001/input/display/geometry
# Response: {"width":1536,"height":776}

---

TL;DR

Extended the server with a new, dedicated set of /input API routes for programmatic control over mouse, keyboard, and window interactions.

Why we made these changes

To introduce comprehensive remote control and automation capabilities by providing a specialized API for interacting with the desktop environment's input devices and windows. This refactors the server to focus specifically on these functionalities.

What changed?

• New API Endpoints: Added /input/mouse/*, /input/keyboard/*, /input/window/*, and /input/display/* routes for mouse movement, clicks, keyboard typing, key presses, and extensive window management (activate, focus, move, resize, close, query information).
• Server Specialization: Refactored main.go to transform the general API server into a dedicated 'input API' server, removing previous OpenAPI specification serving and FFmpeg dependencies.
• New Codebase: Introduced several new Go files (input.go, input_api.go, input_handlers.go, router.go, main_input.go) to define data structures, implement API logic, set up HTTP handlers (often leveraging xdotool), and configure the new server's entry point.
• Documentation: Updated server/README.md with detailed examples and curl commands for all new /input endpoints.
• Testing: Added input_test.go with unit tests for various input handlers.
• Health Check: Included a new /health endpoint to report service status and uptime.

^{Description generated by Mesa. Update settings}

[mesa-dot-dev]

Performed full review of fe02e69...686514f

^{13 files reviewed | 0 comments | Review on Mesa | Edit Reviewer Settings}

[cursor]

Bug: API Endpoints Missing After Update

The server's core recording API endpoints (e.g., /recording/start, /recording/stop) and OpenAPI spec serving (/spec.yaml, /spec.json) are no longer registered. This removes significant existing functionality, as the server now only exposes input API handlers.

[mesa-dot-dev]

Performed full review of fe02e69...686514f

^{13 files reviewed | 0 comments | Review on Mesa | Edit Reviewer Settings}