fix(system-info): tidy up x86 Device Info (CEC state + device model)

[vpetersson]

Summary

Three fixes to the Display & hardware section of the System Info page as it renders on x86 players, reviewed against the x86 testbed.

1. Display Power (CEC) showed "CEC error"

x86 (and any host without a CEC adapter — no /dev/cec0 / /dev/vchiq, e.g. Pi 5) can only fail the libcec probe, so the periodic get_display_power task stored 'CEC error', which the System Info card and the v2 /info API surfaced verbatim — reading like a fault when CEC simply isn't present.

Now short-circuits on the existing diagnostics.cec_available() gate (the same one the Settings page and the display-power SET endpoint use): records a clear Not available and skips spawning a doomed subprocess every tick. CEC-capable boards (Pi 1–4) still report live On/Off.

2 & 3. Device model: stutter + awkward wrap

The card joined the raw DMI sys_vendor + product_name + CPU brand into one string with a ·, so the x86 label both stuttered and word-wrapped at an arbitrary point (splitting the CPU model, e.g. Intel Celeron | 4205U …):

Before

Device model
Intel Corporation Whiskey Platform · Intel Celeron
4205U @ 1.80GHz

After

Device model
Whiskey Platform
Intel Celeron 4205U @ 1.80GHz

• Trailing corporate suffixes (Corporation / Inc. / Ltd. / …) are trimmed, and the board vendor is dropped when the CPU brand already names it (whitebox/reference boards set sys_vendor to the CPU maker). Branded OEM boxes keep their vendor because it differs: Dell Inc. OptiPlex 7090 → Dell OptiPlex 7090.
• get_device_model_parts() now returns (board, cpu) and the template stacks the CPU as the card's secondary line — no ·, no mid-model wrap. Pi keeps its single firmware Model line.

Tests

• get_display_power: existing True/False/error cases run with cec_available() == True; new test asserts 'Not available' is written and the probe skipped when no CEC adapter is present; soft-limit test updated.
• get_device_model_parts: Pi, x86-with-DMI, redundant-vendor drop (Intel reference board), kept-but-trimmed branded vendor (Dell), virtual-chassis drop, and generic fallback — all asserting the (primary, secondary) tuple.
• test_system_info_renders covers the template with the new device_model_detail line.

Verified on the x86 testbed

cec_available() returns False and the card renders "Not available"; the device-model helper produces ('Whiskey Platform', 'Intel Celeron 4205U @ 1.80GHz') for the box's actual DMI values.

🤖 Generated with Claude Code

[Copilot]

Pull request overview

This PR adjusts the periodic get_display_power Celery task so hosts without a CEC adapter (e.g., x86) record "Not available" instead of surfacing "CEC error" via the System Info page and the v2 /info API.

Changes:

• Short-circuit get_display_power on diagnostics.cec_available() and write "Not available" to Redis when CEC isn’t present.
• Update existing tests to force cec_available() == True when exercising the probe path.
• Add a new test asserting the non-CEC path writes "Not available" and does not call the probe.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
src/anthias_server/celery_tasks.py	Short-circuits the display-power periodic task when CEC is unavailable, writing "Not available" to Redis.
tests/test_celery_tasks.py	Adds/updates tests to cover both CEC-present probe behavior and the new CEC-absent short-circuit behavior.

Comments suppressed due to low confidence (1)

src/anthias_server/celery_tasks.py:347

• SoftTimeLimitExceeded is only caught around the probe path; in the non-CEC short-circuit branch the Redis set() can still be interrupted by the soft time limit (or hang until the hard limit), which defeats the task’s “skip the tick instead of SIGKILLing the worker” safety behavior. Consider wrapping the whole body (including the early return) in the same try/except, and make the warning message not specific to the CEC probe so it stays accurate for both branches.

    if not diagnostics.cec_available():
        r.set('display_power', 'Not available', ex=3600)
        return
    try:
        # Single SET with ex= so the value and its TTL are written
        # atomically — a soft-limit signal landing between a separate
        # SET and EXPIRE would otherwise leave the key without a TTL
        # (a stale display_power that never expires).
        r.set('display_power', str(diagnostics.get_display_power()), ex=3600)
    except SoftTimeLimitExceeded:
        # The CEC query is meant to be bounded by its own
        # subprocess timeout, but a child wedged in libcec can keep
        # the pipe open past it. Skip this tick rather than let the
        # hard limit SIGKILL the worker (ANTHIAS-A / 9 / B); the next
        # beat tick re-queries.
        logging.warning(
            'get_display_power: CEC query exceeded %ss; skipping this tick',
            PERIODIC_POKE_SOFT_TIME_LIMIT_S,
        )

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated no new comments.

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[Copilot]

Pull request overview

Copilot reviewed 6 out of 6 changed files in this pull request and generated no new comments.