Cleanup follow-ups from PR #18

[CaseyRo]

Follow-ups from the #18 review. None of these are bugs — #18 is merged and shipped as
v1.4.0-beta.1. These are
complexity cleanups in custom_components/bosch/pointtapi_entities.py, listed roughly by payoff.

Two were already applied before the merge in bd6ee64 (the unreachable /programs/list branch
and a dead _attr_options = []), so they are not repeated here.

1. Number descriptions: two patterns for one problem

_pointtapi_number_descriptions re-indented the whole static NumberEntityDescription tuple
into a function body so two conditional entries (/energy/gas/annualGoal,
/energy/electricity/annualGoal) could be appended. ~120 lines of churn for two appends, and
any future edit to that tuple now conflicts against a reindent.

_pointtapi_select_descriptions solves the identical problem six lines long:

def _pointtapi_select_descriptions(data=None):
    data = data or {}
    return POINTTAPI_SELECT_DESCRIPTIONS + _pointtapi_zone_program_select_descriptions(data)

Same shape works for numbers: keep POINTTAPI_NUMBER_DESCRIPTIONS static, return it plus the
optional goals. The select version is the better one — worth making both look like it.

2. Two on/off parsers in one file

_appliance_error_flag is _resolve_on_off (further down the same file) plus int/float and
the "1"/"0"/"yes"/"no" spellings. Fold those two cases into _resolve_on_off and drop
_appliance_error_flag.

3. _appliance_status_available reads as a wrapped or chain

return any(
    _val(data, f"/system/appliance/{k}") is not None
    for k in ("displayCode", "causeCode", "blockingError", "lockingError")
)

4. _appliance_status_state and _appliance_status_attributes recompute the same six lookups

Both call _appliance_display_code, _appliance_cause_code and both error-flag/code helpers with
identical arguments, on every coordinator update. One _appliance_codes(data) helper, called by both.

---

Deliberately not on this list: the options_fn / current_option_fn / option_to_value_fn
hooks look like abstraction for a single implementation, but sensor descriptions in this file
already carry value_fn / available_fn / attributes_fn / device_info_fn — it is the house
pattern, and a dedicated subclass would not be shorter. The ~100-entry status table is data, not
code. The duplicate-label guard in _zone_program_option_map prevents a program being silently
dropped from the select; it stays.

@jfhautenauven no rush on any of this, and no obligation — happy to take it myself if you would
rather stay on features.

[jfhautenauven]

Thanks for opening this cleanup follow-up @CaseyRo — here is a direct mapping of #19 against what is now delivered through PR #20.

Coverage of requested cleanup items

1) Number descriptions pattern consistency

Requested: align number descriptions with the select pattern (static tuple + dynamic append function).

Delivered in PR #20:

• Number descriptions now use the same shape as selects:
  • static POINTTAPI_NUMBER_DESCRIPTIONS
  • dynamic _pointtapi_dynamic_number_descriptions(data)
  • composed in _pointtapi_number_descriptions(data)
• This removes the large tuple-in-function churn pattern and keeps future edits cleaner.

2) Duplicate on/off parsers

Requested: fold extra bool spellings into _resolve_on_off and drop redundant parser logic.

Delivered in PR #20:

• _resolve_on_off handles broader spellings (on/off, true/false, yes/no, numeric 0/1, numeric active flags).
• Appliance error/status paths now use the unified resolver behavior consistently.

3) _appliance_status_available readability

Requested: keep it as a clean any(...) over appliance keys.

Delivered in PR #20:

• The helper is now in the concise, wrapped any(...) form, using the expected key set.

4) Duplicate appliance-status lookups

Requested: avoid recomputing the same six values in both state and attributes helpers.

Delivered in PR #20:

• Shared _appliance_codes(data) helper introduced.
• Both _appliance_status_state(...) and _appliance_status_attributes(...) consume that shared extraction.

Extra improvements included in the same PR

Beyond #19 cleanup, PR #20 also ships additional quality/value work:

• Thermostat-valve expansion:
  • child lock support
  • actual valve temperature
  • valve position support across compact/full layouts
  • valve temperature offset control (/devices/deviceX/etrv/offset) with min/max/step handling
• Discovery robustness:
  • supports both /devices/list and /devices/deviceN/... structures
• Translation/UX polish:
  • harmonized labels and state wording across locales
  • reduced duplicate/redundant surfaces
• README refresh:
  • richer status badges
  • comprehensive, code-grounded entity inventory and mapping by device/platform/path.

So overall: #19 cleanup goals are covered, and PR #20 also bundles a set of practical extras that should improve maintainability and user-facing clarity.

[CaseyRo]

Leftover from #20's cache removal (63b80d9): unittests/test_pointtapi_setup.py:62 and :95 still define async_load_persistent_cache on the fake coordinators — the real method is gone, so the stubs are dead. Delete both when next in that file.

[jfhautenauven]

@CaseyRo : I think this can be closed with the latest changes on PR #23

[CaseyRo]

Agreed — items 1–4 landed in #20 (_appliance_codes, the any() form, _appliance_error_flag gone, number descriptions composed like the selects), and #23 removes the last two dead stubs. Closing this the moment #23 merges.

[CaseyRo]

All four items shipped in #20; the last dead stubs went with #23 (merged). Thanks @jfhautenauven 🙏