enocean-async
A light-weight, asynchronous, fully typed Python library for communicating with EnOcean devices over a USB gateway. Based on serialx and the EnOcean Serial Protocol Version 3 (ESP3).
Note: The API may still change (even significantly!). Feedback and contributions are welcome.
Features
Receive pipeline — observables
Incoming radio telegrams are decoded into typed Observation objects. Callbacks are available at every stage for lower-level access:
# Stage 4 — semantic: one Observation per entity per device
gateway.add_observation_callback(lambda obs: print(obs))
# Observation(device=…, entity='temperature', values={Observable.TEMPERATURE: 21.3}, …)
# Stage 3 — decoded EEP message (field values + semantic entities)
gateway.add_eep_message_received_callback(lambda msg: ..., sender_filter=eurid)
# Stage 2 — parsed ERP1 telegram (RORG, sender, raw payload bits)
gateway.add_erp1_received_callback(lambda erp1: ...)
# Stage 1 — raw ESP3 packet (before any parsing)
gateway.add_esp3_received_callback(lambda pkt: ...)
Observable members are stable string constants (Observable.TEMPERATURE, Observable.ILLUMINATION, Observable.SWITCH_STATE, Observable.POSITION, Observable.COVER_STATE, Observable.ENERGY, Observable.POWER, Observable.GAS_VOLUME, Observable.GAS_FLOW, Observable.WATER_VOLUME, Observable.WATER_FLOW, …). Each member carries its native unit as Observable.TEMPERATURE.unit == "°C".
Send pipeline — typed instructions
Instructions are sent to devices using typed Instruction subclasses:
from enocean_async import SetCoverPosition, StopCover, SetSwitchOutput
await gateway.send_command(destination=device_eurid, command=SetCoverPosition(position=75))
await gateway.send_command(destination=device_eurid, command=StopCover())
await gateway.send_command(destination=device_eurid, command=SetSwitchOutput(state="on"))
Device management
from enocean_async import device_type_for_eep, EEP, EURID
# Register by EEP — device_type_for_eep() looks up the generic catalog entry
gateway.add_device(address=EURID("01:23:45:67"),
device_type=device_type_for_eep(EEP("D2-05-00")),
name="Living room blind")
# Or register a known manufacturer-specific product from the catalog
from enocean_async import DEVICE_TYPES
nodon_shutter = DEVICE_TYPES["NODON/SIN-2-RS-01"]
gateway.add_device(address=EURID("01:23:45:67"), device_type=nodon_shutter)
device_type_for_eep(eep) raises KeyError for unsupported EEPs. DEVICE_TYPES is a dict[str, DeviceType] keyed by DeviceType.id, containing generic entries (one per supported EEP, manufacturer=None) and manufacturer-specific entries (known physical products). Each DeviceType has a stable id string in NAMESPACE/CODE format (e.g. "EEP/D2-05-00", "NODON/SIN-2-RS-01").
Per-device configuration
Devices support per-device runtime config values (brightness limits, ramp time, etc.). Defaults are populated automatically from the EEP spec at add_device() time and can be overridden at registration or updated later:
# Override at registration time (other keys keep EEP defaults)
gateway.add_device(address=eurid, device_type=...,
config={"min_brightness": 20.0, "max_brightness": 80.0})
# Update at runtime
gateway.set_device_config(eurid, "ramp_time", 5)
Config values are automatically applied in the send path (encoders) and the receive path (semantic resolvers).
EURID, BaseAddress, and Address all accept an int, a colon-separated hex string ("01:23:45:67"), or a 4-byte sequence (bytes, bytearray, list[int]). Use int(addr) and str(addr) for numeric/string conversion.
Learning / teach-in
from enocean_async import TaughtInDevice
def on_taught_in(device: TaughtInDevice) -> None:
print(f"New device: {device.address} ({device.eep})")
gateway.add_device_taught_in_callback(on_taught_in)
await gateway.start_learning(timeout=30)
# gateway now accepts teach-in telegrams and auto-registers devices
gateway.stop_learning()
Supported teach-in methods:
- UTE: automatic bidirectional response; sender address auto-allocated from the base ID pool
- 4BS with profile: auto-registered when EEP is supported; bidirectional response always sent
- Outbound
TeachIn: for Eltako-style sender-addressed actuators, callawait gateway.send_command(address, TeachIn())to send a fixed 4BS payload that registers the gateway's sender slot with the device
1BS teach-in is intentionally not auto-registered (no EEP information available). The NewDeviceCallback fires in all cases.
Focused learning mode: pass focus_device=eurid to start_learning() (or use ToggleLearning(for_device=eurid)) to restrict the learning window to a single device EURID — useful for re-commissioning a specific device from a device config page without accidentally registering nearby devices.
See TEACHIN.md for the full teach-in and teach-out behavior.
Gateway utilities
- Retrieve EURID, Base ID and firmware version info
- Change the Base ID
- Auto-reconnect: when the serial connection is lost, the gateway retries for up to 1 hour
- Gateway device: the gateway itself is observable via
gateway.gateway_entities. Available entities:connection_status—"connected"/"disconnected"/"reconnecting"telegrams_received/telegrams_sent— counters (never reset on reconnect)learning_active—Truewhile a learning session is openlearning_remaining— seconds remaining in the current learning window (counts down per second)learning_toggle— trigger; acceptsToggleLearning()/ToggleLearning(for_device=eurid)learning_timeout— config: default window length in secondslearning_sender— config: sender slot used during teach-in responses
- Per-device
sender_slot: every device gets asender_slotCONFIG_ENUMin itsDeviceSpec.entities. Usegateway.set_device_config(address, "sender_slot", "3")to change it at runtime;device.senderis updated immediately and collisions are checked.
Sender address selection rules
Every outbound telegram carries a sender address. The gateway selects it as follows:
| Device type | Default sender | Reason |
|---|---|---|
Destination-addressed (uses_addressed_sending=True, e.g. D2-01, D2-05) |
BaseID+0 (the base ID itself) | Device is addressed by EURID in the destination field; the sender is irrelevant to routing |
Sender-addressed (uses_addressed_sending=False, e.g. A5-38-08, A5-7F-3F Eltako) |
Next free BaseID+n slot (1–127) | Device filters incoming telegrams by the sender address it learned at teach-in time |
The slot is allocated at teach-in time (or at add_device() time if sender=None) and stored in device.sender. The sender_slot config entity reflects this as "0"–"127" or "eurid". "auto" means no sender has been assigned yet — the first send_command() or TeachIn() that needs one will allocate the next free slot from the pool and backfill device.sender and device.config["sender_slot"].
DeviceSpec.gateway_entities: for sender-addressed devices that need an inbound teach-in from the device (no fixed teach-in payload),device_spec()populates this list withlearning_toggleandlearning_remainingso integrations can surface them on the device's config page (observed/commanded via the gateway's EURID).
What works
- Full receive pipeline: raw serial bytes → ESP3 → ERP1 → EEP decode → observers →
Observationcallbacks - Full send pipeline: typed
Instruction→EEPHandler.encode()→ ERP1 → ESP3 → serial - Device registration with per-device EEP and observer instantiation
- Learning mode: UTE and 4BS-with-profile teach-in (auto-response, device registration, sender pool allocation); 4BS re-teach-in with EEP change supported; focused learning mode (single-EURID restriction)
- Outbound
TeachInfor Eltako-style sender-addressed actuators DeviceTaughtInCallbackwith EURID + EEP on successful teach-in- Auto-reconnect on connection loss
- EURID, Base ID, firmware version retrieval; Base ID change
- Gateway device: connection status, telegram counters, learning state (active/remaining), learning control entities (see Gateway utilities)
- Per-device
sender_slotconfig entity; runtime slot change updatesdevice.senderwith collision detection DeviceSpec.gateway_entitiesfor gateway-sourced entities rendered on device config pages- Parsing of all EEPs listed in SUPPORTED_DEVICES.md
- Sending instructions for: D2-05-00 (covers), D2-20-02 (fan), A5-38-08 (dim gateway + cover status receive + teach-in), A5-7F-3F Eltako FSB (shutter + teach-in), D2-01 (switches/dimmers)
What is missing / not yet implemented
- ECID sub-dispatch for D2-01 extended commands
- More EEPs (contributions welcome — see IMPLEMENT_EEP.md for the step-by-step guide)
- Logging coverage is partial
Implemented EEPs
See SUPPORTED_DEVICES.md.
Architecture
Receive pipeline (observables)
Radio signal
│ serial bytes
▼
EnOceanSerialProtocol3
│ ESP3 framing (sync, CRC, packet type)
▼
ESP3Packet
│ RADIO_ERP1 detection
▼
ERP1Telegram rorg, sender EURID, raw payload bits, rssi
│ EEP profile lookup → EEPHandler.decode()
▼
EEPMessage
.values {field_id → EEPMessageValue} ← EEP spec vocabulary: "TMP", "ILL1", "R1"
.entities {observable → EntityValue} ← semantic vocabulary: TEMPERATURE, ILLUMINATION
│ Observer.decode() (one call per observer in device.observers)
├── ScalarObserver(observable=TEMPERATURE) → reads entities[TEMPERATURE]
├── ScalarObserver(observable=ILLUMINATION) → reads entities[ILLUMINATION]
├── CoverObserver → reads entities[POSITION] + entities[ANGLE], infers COVER_STATE
├── ButtonObserver → reads values["R1"], values["EB"], … (stateful, hold timer)
└── MetaDataObserver → emits rssi, last_seen, telegram_count
│ _emit()
▼
Observation(device, entity, values, timestamp, source)
│ add_observation_callback
▼
Application
Send pipeline (instructions)
Application
│ gateway.send_command(destination, command=SetCoverPosition(position=75))
▼
Instruction subclass (typed dataclass with ClassVar[Instructable] action)
│ spec.encoders[command.action](command, device.config)
▼
EEPMessage
.message_type ← selects which telegram type to encode
.values ← {field_id → EEPMessageValue(raw)} filled in by the encoder
│ EEPHandler.encode()
├── Determine buffer size from field layout
├── Write CMD bits at cmd_offset / cmd_size
└── Write each field's raw value at field.offset / field.size
▼
ERP1Telegram(rorg, telegram_data, sender, destination)
│ .to_esp3()
▼
ESP3Packet
│ Gateway.send_esp3_packet()
▼
Radio signal → Device
See ARCHITECTURE.md for a detailed description of the EEP layer, the semantics layer, and the key design decisions.
Contributing
See CONTRIBUTING.md.
Versioning
See VERSIONING.md for the version scheme and bump instructions.
Dependencies
This library has one dependency:
Technology documentation
- EnOcean Serial Protocol Version 3 (ESP3)
- EnOcean Radio Protocol 1 (ERP1)
- EnOcean Alliance Specifications
- EURID Specification V1.2
- EEP V3.1 (high-level)
- EEPViewer (individual profiles)
Copyright & license
Copyright 2026 Henning Kerstan
Licensed under the Apache License, Version 2.0 (the "License"). See LICENSE file for details.
Release files for enocean-async 0.16.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| enocean_async-0.16.0.tar.gz | 100.3 kB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| enocean_async-0.16.0-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 215.8 kB
Release files / enocean_async-0.16.0.tar.gz
| Download URL | enocean_async-0.16.0.tar.gz |
|---|---|
| Size | 100.3 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
e468c0d44e06f4f544a98762c04eba0a38b38771245919b55fac2ff3108347e4
|
|
BLAKE2b-256 checksum How to use checksums |
4f05b735fcd384dfb9e5bb5ba2f9378ceb5fa2fa25737b05997670c79b14190a
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.12
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on May 28, 2026.
Transparency logRelease files / enocean_async-0.16.0-py3-none-any.whl
| Download URL | enocean_async-0.16.0-py3-none-any.whl |
|---|---|
| Size | 115.5 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
6dfeea0ee2920f74a65af1fd0e011403c02809683ae54b78fbc5c47562c73bb4
|
|
BLAKE2b-256 checksum How to use checksums |
e0ae5d7892363dc95796aba73642989f6942f40a4b90dd7e5da8f723172b76bc
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.12
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on May 28, 2026.
Transparency log