docs: refactor docs structure and tighten assistant instruction policy

shrink root README into a landing page with a docs map and focused contributor guidance
add TV_POWER_RUNBOOK as the canonical TV power rollout and canary runbook
add CHANGELOG and move project history out of README-style docs
refactor src README into a developer-focused guide (architecture, runtime files, MQTT, debugging)
prune redundant older HDMI docs and keep a canonical HDMI_CEC_SETUP path
update copilot instructions to a high-signal policy format with strict anti-shadow-README design rules
align references across docs to current files, scripts, and TV power behavior
This commit is contained in:
RobbStarkAustria
2026-04-01 10:01:58 +02:00
parent fb0980aa88
commit 82f43f75ba
20 changed files with 2228 additions and 2267 deletions

View File

@@ -0,0 +1,163 @@
# TV Power Intent — Server Contract v1 (Phase 1)
> This document is the stable reference for client-side implementation.
> The server implementation is validated and frozen at this contract.
> Last validated: 2026-04-01
---
## Topic
```
infoscreen/groups/{group_id}/power/intent
```
- **Scope**: group-level only (Phase 1). No per-client topic in Phase 1.
- **QoS**: 1
- **Retained**: true — broker holds last payload; client receives it immediately on (re)connect.
---
## Publish semantics
| Trigger | Behaviour |
|---|---|
| Semantic transition (state/reason changes) | New `intent_id`, immediate publish |
| No change (heartbeat) | Same `intent_id`, refreshed `issued_at` and `expires_at`, published every poll interval |
| Scheduler startup | Immediate publish before first poll wait |
| MQTT reconnect | Immediate retained republish of last known intent |
Poll interval default: **15 seconds** (dev) / **30 seconds** (prod).
---
## Payload schema
All fields are always present. No optional fields for Phase 1 required fields.
```json
{
"schema_version": "1.0",
"intent_id": "<uuid4>",
"group_id": <integer>,
"desired_state": "on" | "off",
"reason": "active_event" | "no_active_event",
"issued_at": "<ISO 8601 UTC with Z>",
"expires_at": "<ISO 8601 UTC with Z>",
"poll_interval_sec": <integer>,
"active_event_ids": [<integer>, ...],
"event_window_start": "<ISO 8601 UTC with Z>" | null,
"event_window_end": "<ISO 8601 UTC with Z>" | null
}
```
### Field reference
| Field | Type | Description |
|---|---|---|
| `schema_version` | string | Always `"1.0"` in Phase 1 |
| `intent_id` | string (uuid4) | Stable across heartbeats; new value on semantic transition |
| `group_id` | integer | Matches the MQTT topic group_id |
| `desired_state` | `"on"` or `"off"` | The commanded TV power state |
| `reason` | string | Human-readable reason for current state |
| `issued_at` | UTC Z string | When this payload was computed |
| `expires_at` | UTC Z string | After this time, payload is stale; re-subscribe or treat as `off` |
| `poll_interval_sec` | integer | Server poll interval; expiry = max(3 × poll, 90s) |
| `active_event_ids` | integer array | IDs of currently active events; empty when `off` |
| `event_window_start` | UTC Z string or null | Start of merged active coverage window; null when `off` |
| `event_window_end` | UTC Z string or null | End of merged active coverage window; null when `off` |
---
## Expiry rule
```
expires_at = issued_at + max(3 × poll_interval_sec, 90s)
```
Default at poll=15s → expiry window = **90 seconds**.
**Client rule**: if `now > expires_at` treat as stale and fall back to `off` until a fresh payload arrives.
---
## Example payloads
### ON (active event)
```json
{
"schema_version": "1.0",
"intent_id": "4a7fe3bc-3654-48e3-b5b9-9fad1f7fead3",
"group_id": 2,
"desired_state": "on",
"reason": "active_event",
"issued_at": "2026-04-01T06:00:03.496Z",
"expires_at": "2026-04-01T06:01:33.496Z",
"poll_interval_sec": 15,
"active_event_ids": [148],
"event_window_start": "2026-04-01T06:00:00Z",
"event_window_end": "2026-04-01T07:00:00Z"
}
```
### OFF (no active event)
```json
{
"schema_version": "1.0",
"intent_id": "833c53e3-d728-4604-9861-6ff7be1f227e",
"group_id": 2,
"desired_state": "off",
"reason": "no_active_event",
"issued_at": "2026-04-01T07:00:03.702Z",
"expires_at": "2026-04-01T07:01:33.702Z",
"poll_interval_sec": 15,
"active_event_ids": [],
"event_window_start": null,
"event_window_end": null
}
```
---
## Validated server behaviours (client can rely on these)
| Scenario | Guaranteed server behaviour |
|---|---|
| Event starts | `desired_state: on` emitted within one poll interval |
| Event ends | `desired_state: off` emitted within one poll interval |
| Adjacent events (end1 == start2) | No intermediate `off` emitted at boundary |
| Overlapping events | `desired_state: on` held continuously |
| Scheduler restart during active event | Immediate `on` republish on reconnect; broker retained holds `on` during outage |
| No events in group | `desired_state: off` with empty `active_event_ids` |
| Heartbeat (no change) | Same `intent_id`, refreshed timestamps every poll |
---
## Client responsibilities (Phase 1)
1. **Subscribe** to `infoscreen/groups/{own_group_id}/power/intent` at QoS 1 on connect.
2. **Re-subscribe on reconnect** — broker retained message will deliver last known intent immediately.
3. **Parse `desired_state`** and apply TV power action (`on` → power on / `off` → power off).
4. **Deduplicate** using `intent_id` — if same `intent_id` received again, skip re-applying power command.
5. **Check expiry** — if `now > expires_at`, treat as stale and fall back to `off` until renewed.
6. **Ignore unknown fields** — for forward compatibility with Phase 2 additions.
7. **Do not use per-client topic** in Phase 1; only group topic is active.
---
## Timestamps
- All timestamps use **ISO 8601 UTC with Z suffix**: `"2026-04-01T06:00:03.496Z"`
- Client must parse as UTC.
- Do not assume local time.
---
## Phase 2 (deferred — not yet active)
- Per-client intent topic: `infoscreen/{client_uuid}/power/intent`
- Per-client override takes precedence over group intent
- Client state acknowledgement: `infoscreen/{client_uuid}/power/state`
- Listener persistence of client state to DB