0e9a353d75
controller: jinja2 template для chat overlays + UI marker для icon/text
...
dynamic_overlays.ChatCfg.template (опц.) — Jinja2 шаблон применяется к
incoming MQTT payload. Доступны переменные payload (raw str) и payload_json
(parsed JSON dict если valid). Используется для извлечения отдельного поля
из JSON, например temperature из z2m sensor.
Применение в controller.yaml:
chats:
- id: temp_outside
source_topic: "zigbee2mqtt/Температура на улице"
template: "{{ '%+.1f' | format(payload_json.temperature | float) }}°C"
Если template не задан — payload используется как есть (backwards compat
с alerts_chat и подобными).
UI editor static/index.html: для icon/text overlays вместо коробки показываем
точечный маркер 14×14 (синий circle). Причина: filter render_overlay_icon
использует native PNG dimensions, frontend не знает реальный размер
(w_px/h_px хранятся в ChatCfg на server-side). Показ size=10% как было —
вводил в заблуждение. Resize handle тоже скрыт для marker'ов.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com >
2026-05-26 20:55:35 +01:00
19ddaf2dde
controller: browser-rendered overlays — Grafana/chat/любой HTML
...
MVP Phase 8 feature: headless Chromium snapshot URL → PNG → existing icon
overlay infrastructure. Use case — Grafana dashboard, web chat, любой
HTML widget с transparent background поверх video composite.
Архитектура:
BrowserRenderer launches один shared Chromium instance, per-dashboard
page. Loop:
page.reload (для свежих данных — Grafana auto-refresh не triggers
в headless mode без user interaction)
page.add_style_tag (re-inject transparent CSS — reload teryает styles)
page.wait_for_selector (если selector задан)
page.screenshot(omit_background=True) или locator(selector).screenshot
save PNG к icon_dir
dispatcher._reload_icon → filter re-reads atlas
Config (dynamic_overlays.dashboards в controller.yaml):
id, target_instance, cell, url, x, y, w_px, h_px,
refresh_sec (default 2.0s, min 0.5s — frequent reload bad см. stutter memory),
inject_css (default — transparent background + zero margins),
selector (optional CSS selector — экспортить только конкретный element),
viewport_w/h (override page viewport, default = w_px/h_px;
полезно при selector чтобы page layout не collapsed),
wait_until, page_timeout_ms, opacity, z_order.
Dockerfile:
+ pip install playwright
+ playwright install --with-deps chromium (~170MB Chromium + ~150MB system libs)
Если dashboards не нужны — можно убрать оба install'а (playwright
import lazy, BrowserRenderer.start() graceful no-op без playwright).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com >
2026-05-21 23:09:50 +01:00
d807cd2c23
controller: Phase 5b+5c+6 — multi-audio + intercom ducking + dynamic overlays
...
5b — audio source switching:
AudioSourceCfg list + audio_filter_target в InstanceCfg
CommandDispatcher._audio_set → ZMQ astreamselect@as map <index>
REST: GET /audio/{inst}, POST /audio/{inst}/set
MQTT: cuda_grid/cmd/<inst>/audio/set <source_name>
5c — intercom ducking:
music_volume_target / intercom_volume_target / music_ducked_volume в InstanceCfg
CommandDispatcher._intercom_set → 2× ZMQ volume@music/@intercom commands
REST: POST /intercom/{inst}/start (music↓ + intercom↑) + /end (restore)
MQTT: cuda_grid/cmd/<inst>/intercom/start|end
6 — dynamic overlays (charts/chats):
dynamic_overlays.py: ChartCfg/ChatCfg + DynamicRenderer
PIL rendering: line chart + scrolling text list
Async loops пишут PNG в icon_dir + invalidate filter cache via reload_icon ZMQ
MQTT subscriptions для real data (charts: numeric topic, chats: text topic)
Demo: chart sine wave если data_topic=null
Wired в __main__.py + mqtt_loop dispatch
+ ZMQ client asyncio.Lock — REQ socket strict send/recv pattern требует
serialize requests (overlay/audio/intercom concurrent ломали "Operation
cannot be accomplished in current state")
+ Pillow в Dockerfile (для PIL render)
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com >
2026-05-20 21:55:33 +01:00
37232ae1b9
controller: Phase 3 — Python sidecar skeleton (MQTT + ZMQ + HTTP + HA Discovery)
...
cuda-grid-controller (Python 3.11+) — control plane между HA/MQTT/HTTP
и FFmpeg's vf_cuda_grid filter через ZMQ.
Modules (~700 LOC Python):
- config.py — Pydantic schema (broker, instances[], ha_discovery, http, log) + YAML loader
- layouts.py — registry известных layouts (sync с vf_cuda_grid.c Phase 2)
- ha_discovery.py — HA MQTT Discovery payloads (select.layout, sensor.current_layout,
binary_sensor.online per instance + global device entry)
- zmq_client.py — async ZMQ REQ socket к FFmpeg zmq filter
(target command args → reply parsing)
- state.py — in-memory ControllerState (active_layout per instance, asyncio.Lock)
- mqtt_loop.py — aiomqtt async loop: subscribe cuda_grid/cmd/<inst>/+/+,
publish cuda_grid/state/* (retained) + cuda_grid/event/*, LWT, HA status reconnect
- dispatch.py — CommandDispatcher: layout.set action → ZMQ send_command + state update + events
- http_api.py — FastAPI: /health, /layouts, /state, POST /layout/{inst}/set
- __main__.py — typer CLI, asyncio.gather(mqtt_loop, uvicorn.server)
Examples + Dockerfile:
- examples/controller.yaml — 2 instances (livingroom_tv, public_stream)
- Dockerfile — python:3.11-slim, ENTRYPOINT cuda-grid-controller
- README — overview, usage, FFmpeg side filter graph
End-to-end flow ready:
HA dashboard → MQTT → controller → ZMQ → FFmpeg process_command → layout switch
↓
state публикуется обратно в MQTT → HA UI обновляется
Phase 3 deliverable per gx/vf-cuda-grid#1 . Phase 4 = overlays (rect/text/icon).
2026-05-19 21:52:11 +01:00
gx