Replaces the read-only "structured comparison" banner with a real
editor. Structured AND records encode ``Arg1 OP Arg2`` where Arg1 is
a typed reference (Zone / Unit / Thermostat / Area / TimeDate) plus a
per-type field selector, and Arg2 is either another typed reference
or a literal constant.
I1 — TS types + decoders:
Wire layout (programs.py decoders, clsProgram.cs):
cond high byte = and_op (CondOP: 1=EQ, 2=NE, 3=LT,
4=GT, 5=ODD, 6=EVEN, 7=MULT,
8=IN, 9=NOT_IN)
cond low byte = and_arg1_argtype (CondArgType)
cond2 (whole) = and_arg1_ix (object idx; 0 for TimeDate)
cmd = and_arg1_field (per-type field selector)
par = and_arg2_argtype (Constant most common)
pr2 = and_arg2_ix (constant value or 2nd obj idx)
month = and_arg2_field
day,days = and_compconst (BE u16; usually 0)
decodeStructuredAnd / encodeStructuredAnd handle both directions;
round-trip exact.
Per-Arg1Type field menus in FIELDS_BY_TYPE — exact 1:1 with the
Python enuZoneField / enuUnitField / enuThermostatField /
enuTimeDateField enums in omni_pca.programs and the field handling
in StateEvaluator. Areas only expose "Security mode" (single useful
field). TimeDate exposes Year / Month / Day / DoW / Time / Hour /
Minute (skips the rarely-used Date / DST / SunriseSunset fields).
I2 — editor UI:
isEditableStructuredAnd guard: only opens the editor for records
matching the editor's scope (Arg1 in supported types, Arg2=Constant,
compConst=0). Out-of-scope structured records render with a
"read-only" tag — preserved on save, still removable.
Structured rows render with a "structured" tag and an orange-tinted
background to distinguish them from Traditional rows. Layout:
Arg1 type ▸ object picker ▸ Field ▸ Operator ▸ Compare against
Unary operators (ODD / EVEN) hide the Arg2 input. Changing Arg1 type
resets the Arg1 index + field to defaults so the form stays self-
consistent (no stale picker values from a previous type).
Arg2 is locked to Constant in this pass. Editing record-vs-record
comparisons (e.g. "Thermostat 1 temp > Thermostat 2 temp") is a
future cut — current real-world programs use the Constant form
exclusively per my homeowner-panel sample.
_pickBucket gains the missing "thermostat" branch (was missed in
earlier passes; only mattered now that thermostat is an Arg1Type).
Live screenshot 12-structured-and.png shows an injected chain with
both a Traditional AND (CTRL UNIT 1 ON) and a Structured AND
(Thermostat(1).Temperature > 70) — both editable end-to-end.
Frontend bundle: 88 KB minified (up from 82 KB).
Full suite: 653 passed, 1 skipped (no test changes).
Multi-record clausal programs are now editable end-to-end. A chain
spans N consecutive slots — head (WHEN/AT/EVERY) + zero-or-more
AND/OR condition records + one-or-more THEN action records — so
"editing" means rewriting the whole run, validating that any
expansion doesn't trample adjacent programs, and clearing any old
slots when the chain shrinks.
H1 — backend:
* programs/get for chains now returns chain_members[] with each
member's slot + role + raw fields. The editor uses this to seed
one editable form-row per slot.
* New programs/chain/write command: takes head_slot + head dict +
conditions[] + actions[], does N sequential download_program
calls, then clears any old chain slots that fell outside the new
range. Validates:
- head_slot + new_len doesn't extend past slot 1500
- any expansion-into slot not already part of THIS chain is FREE
(anti-trample: refuse rather than overwrite an adjacent program)
- at least one THEN action present (empty chain rejected)
Updates coordinator.data.programs immediately so subsequent list
calls reflect the edit before the next poll.
H2 — TS helpers:
* AND-record encoding mirrors compact-form cond family bytes
(0x04 ZONE / 0x08 CTRL / 0x0C TIME / 0x00 OTHER + 0x10+ SEC) but
with a slightly different bit layout: the family byte lives at
fields.cond & 0xFF (disk byte 1) and the instance at
(fields.cond2 >> 8) & 0xFF (disk byte 3). The selector bit is
family's bit 0x02 instead of cond's 0x0200. decodeAndCondition /
encodeAndCondition handle both directions; round-trip exact.
* isStructuredAnd helper detects records with OP > 0 (TEMP > N
comparisons etc.); those render read-only in the chain editor
with a warning banner.
* emptyAndRecord / emptyOrRecord / emptyThenRecord helpers for
the add-condition / add-action buttons.
H3 — chain editor UI:
* New _chainDraft state (parallel to _editingDraft for compact form)
with head + conditions[] + actions[] arrays. Mutation helpers
preserve immutability via array-copy-then-patch.
* "Edit" button on chain detail now opens the chain editor instead
of returning early (previous read-only behaviour).
* Three sub-renderers: trigger section dispatches on prog_type
(WHEN→event-id builder reusing the EVENT helpers, AT→time+days
reusing TIMED layout, EVERY→single seconds input that packs into
cond+cond2), conditions section with per-row add/remove (separate
+ AND IF and + OR IF buttons in the legend), actions section with
per-row add/remove (+ THEN button; at least one action enforced).
* Structured-OP AND records render with an explanatory read-only
banner and a × button to drop the row entirely — preserves the
data when the user doesn't touch it, lets them remove it cleanly
when they want to.
* Each row picks objects via _bucketWithPreserve so out-of-range
zone/unit/area indices stay safe.
5 new HA integration tests:
* get-chain returns chain_members with correct roles + raw fields
* chain/write in-place rewrite preserves footprint, updates bytes
* chain/write shrink clears the trailing old slots
* chain/write refuses to trample an adjacent program on expansion
* chain/write rejects zero-actions submission
Live screenshot 11-chain-editor.png: state injection into the side
panel (real panel has no chains) shows the editor rendering a sample
WHEN zone-state → AND IF unit ON → 2x THEN action chain with every
control populated and functional.
Full suite: 653 passed, 1 skipped (up from 648, 5 new chain tests).
Frontend bundle: 82 KB minified (up from 63 KB).
Replaces the read-only "conditions present but not editable" banner
with a real editor for the cond / cond2 u16 fields on TIMED / EVENT /
YEARLY programs.
Compact-form conditions split into five families per
clsText.GetConditionalText (clsText.cs:2224-2274):
none — cond = 0 (no inline condition)
misc — family 0x00, low nibble = enuMiscConditional
(NONE / NEVER / LIGHT / DARK / PHONE_* / AC_POWER_* /
BATTERY_* / ENERGY_COST_*)
zone — family 0x04, low byte = zone, bit 0x0200 = NOT_READY
unit — family 0x08, low 9 bits = unit, bit 0x0200 = ON
time — family 0x0C, low byte = time-clock #, bit 0x0200 = enabled
sec — family >= 0x10, bits 8-11 = area, bits 12-14 = security mode
types.ts gains decodeCondition / encodeCondition + the
MISC_CONDITIONALS / SECURITY_MODE_NAMES enums. Round-trip is exact:
decode(encode(c)) === c for every supported family.
UI: two condition slots per editor (matching the two u16 fields on
the wire). Each slot has a family-picker dropdown that swaps the
sub-fields (zone picker + secure/not-ready, unit picker + on/off,
area picker + security mode, time-clock # + enabled/disabled, misc
condition picker, or "none"). Picking a family seeds sensible defaults
(NEVER for misc, first zone secure, first unit ON, area 1 disarmed,
time clock 1 enabled).
Object pickers reuse the same _bucketWithPreserve helper introduced
for the action editor, so out-of-range zone/unit/area refs in inline
conditions keep their original value with a "preserve" label.
Live smoke test against the real panel: slot #1's actual condition
"AND IF Time clock 4 is disabled" now decodes into the editor as
Family=Time clock / # = 4 / Is = disabled — exactly the on-disk state.
Frontend bundle: 63 KB minified (up from 56 KB with the new editor
section + cond helpers).
Real-world programs reference object indexes well past the coordinator's
discovery range — typical example from a live OmniPro II: a program
that's "Turn ON Unit 33025" where the unit number is a raw byte value
from undecoded extended-output addressing. The discovered units bucket
only covers slots up to ~511, so 33025 doesn't match any entry.
Before this commit the dropdown silently fell through to the first
known unit (e.g. ROOM ONE), making it look like the user had selected
that unit. The underlying draft.pr2 stayed at 33025, but a user who
glanced at the form and clicked Save would either preserve the original
(if they didn't touch the select) or accidentally clobber it with the
first list item (if they did).
Fix: _bucketWithPreserve prepends a synthesized option
"(undiscovered <kind> <idx> — preserve original)" when the current
value isn't represented. Applies in all four picker sites:
* Action object picker (Unit / Zone / Area / Button for action commands)
* EVENT trigger Button picker
* EVENT trigger Zone picker
* EVENT trigger Unit picker
The synthesized entry sits at the top of the list (visually distinct)
and is the selected default. Picking any other entry from the dropdown
then becomes an explicit choice — no more silent coercion.
Smoke-tested against the real panel: slot #1 (WHEN OPEN BIG GAR →
Turn ON Unit 33025) now shows "#33025 (undiscovered unit 33025 —
preserve original)" as the selected Unit option. Screenshots updated.
Two bugs surfaced when smoke-testing against a real OmniPro II:
1. Empty list after page load. _discoverViaList ran fire-and-forget;
connectedCallback then synchronously checked _entryId (still null
because await hadn't resolved) and skipped _loadList. The panel
rendered "No programs match the current filters" forever — until
the next 5-second poll tick, which never fires because
_startRefreshTimer was also gated on the same null check.
Fix: have _discoverViaList itself trigger _loadList and
_startRefreshTimer after _entryId lands. The connectedCallback /
updated paths can stay gated on _entryId; the discover path now
takes ownership of "do the initial load too."
2. Dev installs with both a working entry and a setup_retry entry
(mock container down, real panel up) had the panel pick the
setup_retry one first and surface "panel not configured" on every
call. Fix: prefer entries with state === "loaded" in the discover
step, falling back to first entry only when none are loaded.
Also: screenshot.py drops the seed-via-WS step (was unsafe — would
write Programs to whatever entry is loaded, including real panels).
Updates the in-page click helpers to walk the shadow DOM recursively
instead of hardcoding HA's host-element path, so detail/editor
screenshots work on the actual depth-8 element location.
Smoke test against real panel: 154 programs render correctly with
structured English, BEDTIME / OPEN BIG GAR / Zone 133 events all
decoded, B. GAR MAN DOOR [SECURE] live-state badge visible.
Detail panel + editor mode both function end-to-end.
dev/screenshot.py — end-to-end automated demo:
* onboards HA via /api/onboarding (user creation + auth_code flow)
* subsequent runs log in via /auth/login_flow with saved credentials
* adds the omni_pca config entry via /api/config/config_entries/flow
* uses HA's template REST endpoint to discover the panel device_id
* launches a headless chromium via playwright with prefetched auth tokens
* captures 6 deep-linked screenshots:
01-overview.png — Lovelace
02-integrations-list.png — HAI/Leviton sitting next to HA's built-ins
03-omni-pca-config.png — '1 device · 38 entities', custom integration
04-panel-device.png — Omni Pro II device page with full controls
05-entities-omni.png — config_entry filtered entities table
06-developer-states.png — alarm_control_panel.omni_pro_ii_main with
raw_mode_name=OFF, code_arm_required=true,
etc. proving real entity state from mock
dev/docker-compose.yml — mock-panel command rewritten:
* Mounts only src/ and run_mock_panel.py (read-only) instead of the full
project so uv doesn't try to recreate the host's .venv on a RO mount
* Installs cryptography via uv pip install --system
* PYTHONPATH set to /tmp/mock/src so omni_pca imports work without a
package install
dev/artifacts/screenshots/2026-05-10/ — six PNGs from the run.
.gitignore — adds dist/ for build artifacts.
Confirmed end-to-end: HA discovered the integration via mDNS hint
(showed up in the onboarding wizard's compatible-devices step), the
config flow connected to the mock over host.docker.internal:14369,
materialized 38 entities across 8 platforms (alarm_control_panel,
binary_sensor, button, climate, event, light, sensor, switch), and
displayed everything in the device + entity registries with friendly
names and attributes intact. The integration name hash is 38 entities
because the mock seeds 5 zones (binary + bypass) + 4 units + 2 areas +
2 thermostats + 3 buttons + 3 system-level binary sensors + 2 system
sensors + 6 thermostat sensors + 1 event entity = 38 (matches HA UI).
