Add SwitchbotKeypad device class for classic Keypad

[italo-lombardi]

Adds support for SwitchBot Keypad (WoKeypad) — exposes battery level and attempt_state from BLE advertisements.
https://eu.switch-bot.com/products/switchbot-keypad

Tested on a real device — scanner confirmed battery: 100 and attempt_state parsed correctly from live BLE advertisement data.

Summary

The classic SwitchBot Keypad (WoKeypad) is a passive BLE-only device used to unlock doors paired with SwitchBot Lock. The advertisement parser (adv_parsers/keypad.py) already parses battery and attempt_state from BLE advertisements, and SwitchbotModel.KEYPAD is already defined, but no device class existed.

Changes

• Add SwitchbotKeypad device class in devices/keypad.py, extending SwitchbotDevice (passive-only, no BLE commands)
• Export SwitchbotKeypad from switchbot/__init__.py
• Add KEYPAD_INFO test fixture in tests/__init__.py
• Add tests/test_keypad.py verifying battery and attempt_state are correctly parsed from advertisement data

Test results

All 3 unit tests pass. Verified against real WoKeypad hardware — advertisement data parsed correctly.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

Files with missing lines	Coverage Δ
switchbot/__init__.py	100.00% <100.00%> (ø)
switchbot/devices/keypad.py	100.00% <100.00%> (+100.00%)	⬆️

... and 1 file with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[bdraco]

@bluetoothbot review

[Copilot]

Pull request overview

Adds a minimal SwitchbotKeypad device class that wraps the existing process_wokeypad advertisement parser, exposes it from the package's public API, and adds a fixture plus three tests verifying battery and attempt_state parsing.

Changes:

• New switchbot/devices/keypad.py defining SwitchbotKeypad (subclass of SwitchbotDevice, no commands).
• Public export added in switchbot/__init__.py (SwitchbotKeypad).
• New KEYPAD_INFO AdvTestCase fixture in tests/__init__.py and a new tests/test_keypad.py exercising advertisement parsing.

Reviewed changes

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

File	Description
switchbot/devices/keypad.py	Adds the new passive SwitchbotKeypad device class.
switchbot/init.py	Imports and re-exports SwitchbotKeypad in the public API.
tests/init.py	Adds KEYPAD_INFO advertisement fixture used by the new tests.
tests/test_keypad.py	Adds tests verifying parsed battery and attempt_state values, including battery=None override.

---

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

[bluetoothbot]

PR Review — Add SwitchbotKeypad device class for classic Keypad

Small, well-scoped change that fills a real gap: the process_wokeypad parser and SwitchbotModel.KEYPAD enum already existed, but no device class consumed them. Implementation is correctly minimal (passive BLE — no commands), the public export is alphabetized, and codecov reports 100% coverage on the new code. The author addressed the prior automated review by driving tests through parse_advertisement_data with real BLE bytes and asserting on get_battery_percent() rather than stuffing pre-parsed values — that's the right call and the tests now actually exercise the parser path end-to-end. Author flagged that the code wasn't tested against real hardware, but the byte-layout is the same as the existing test_parse_advertisement_data_keypad fixture in test_adv_parser.py:1450, so regressions would surface. Merge-ready; the inline comments are nits.

---

🟢 Suggestions

1. Repeated setup could be pulled into a fixture or parametrize (`tests/test_keypad.py`, L9-58)

All three tests duplicate the BLE device + advertisement construction. With three short tests it is acceptable as-is, but extracting a small helper or using @pytest.fixture for the parsed advertisement (and @pytest.mark.parametrize for the empty-mfr-data case) would shrink the file and make future additions cleaner.

Not a blocker — the duplication is bounded and the intent is clear.

2. Consider also asserting attempt_state is None in the empty-mfr-data case (`tests/test_keypad.py`, L42-58)

test_keypad_advertisement_battery_none_when_no_data currently only checks that get_battery_percent() is None. Since process_wokeypad returns {"battery": None, "attempt_state": None} together when mfr_data is None (see adv_parsers/keypad.py:15-16), it would be cheap to also assert device.parsed_data["attempt_state"] is None to fully exercise that branch and document the contract.

assert device.get_battery_percent() is None
assert device.parsed_data["attempt_state"] is None

3. Optional: reference SwitchbotModel.KEYPAD in the docstring (`switchbot/devices/keypad.py`, L7-14)

The class body is intentionally empty (correct for a passive BLE device). To help discoverability — especially relative to the very similarly named SwitchbotKeypadVision (which is an encrypted, command-capable device) — consider mentioning SwitchbotModel.KEYPAD and contrasting with the Vision variant in the class docstring. Purely cosmetic.

---

Checklist

• Class follows pattern of other passive device classes
• Public export added in alphabetical order
• Tests exercise the full parse_advertisement_data pipeline
• None-handling branch covered — suggestion ＃2
• No hardcoded secrets, unsafe deserialization, or injection vectors
• Coverage of new code
• Python 3.11+ compatible

---

Summary

Small, well-scoped change that fills a real gap: the process_wokeypad parser and SwitchbotModel.KEYPAD enum already existed, but no device class consumed them. Implementation is correctly minimal (passive BLE — no commands), the public export is alphabetized, and codecov reports 100% coverage on the new code. The author addressed the prior automated review by driving tests through parse_advertisement_data with real BLE bytes and asserting on get_battery_percent() rather than stuffing pre-parsed values — that's the right call and the tests now actually exercise the parser path end-to-end. Author flagged that the code wasn't tested against real hardware, but the byte-layout is the same as the existing test_parse_advertisement_data_keypad fixture in test_adv_parser.py:1450, so regressions would surface. Merge-ready; the inline comments are nits.

---

Automated review by Kōan570b216
7430fd3
72f509b
09dd94e
e2174d0
0d4085c
d0167e8

[italo-lombardi]

Addressed all suggestions from the automated review:

1. Tests now exercise the full pipeline — rewrote all three tests to call parse_advertisement_data with real BLE bytes instead of injecting pre-parsed values. Added assertion on get_battery_percent().

2. Fixed misleading test — test_keypad_advertisement_battery_none_when_no_data now passes empty manufacturer_data so the None comes from the parser's own None-handling branch (adv_parsers/keypad.py:15-16), not a manual override.

3. Removed copy-paste — moved make_advertisement_data to tests/__init__.py as a shared helper. test_keypad_vision.py now imports it from there.

4. Added docstring to SwitchbotKeypad — clarifies passive-only BLE, no commands, and points to adv_parsers/keypad.py for data origin.

5. Fixture bytes — KEYPAD_INFO in tests/__init__.py is already the single source of truth; the new tests reference it directly, so test_adv_parser.py could be updated to do the same in a follow-up if desired.

[italo-lombardi]

@bluetoothbot review

[bluetoothbot]

❌ Permission denied. Only users with write access can trigger bot commands.

[bdraco]

@bluetoothbot review

[andreasbrett]

What's the status here? Would love to have this in the upcoming HA release. Seems like the bot did not fire again.

[bdraco]

The code as-is doesn't do anything. There is nothing to review

[italo-lombardi]

This PR was AI-assisted. The goal is to expose battery % for the classic WoKeypad in Home Assistant.

The pySwitchbot side should be complete: SwitchbotKeypad inherits get_battery_percent() from SwitchbotDevice, which reads the "battery" key that process_wokeypad already parses from the advertisement data. No BLE commands needed — this is a passive device.

A companion HA PR is needed to wire it up (NON_CONNECTABLE_SUPPORTED_MODEL_TYPES + SupportedModels.KEYPAD + CLASS_BY_DEVICE), similar to how LEAK and CONTACT_SENSOR are handled. That PR will come once this merges.

I have the physical device but no experience testing pySwitchbot integrations. Any guidance on how to validate this locally (e.g. with a BLE sniffer or test harness against real hardware) would be appreciated.

[italo-lombardi]

Update: tested on real hardware.

Ran the parser against a live WoKeypad using BLE scanning on macOS.

Results from advertisement data:

• battery: 100%
• attempt_state: 41 (value at time of scan, changes per keypad interaction)

Parser correctly extracts both values from service_data[2] & 0x7f and manufacturer_data[6]. Length guards (>= 3 svc, >= 7 mfr) satisfied by real device advertisements.

All 3 unit tests pass locally.

[CodeFinder2]

Really cool, would love to see this merged! :-)