cez-distribution-hdo

Python library for reading and interpreting HDO (low/high tariff) switch times from the CEZ Distribution “switch-times / signals” API.

This repository contains the core library only. A Home Assistant integration is planned as a separate project.

Features

• Async HTTP client (httpx) for the CEZ Distribution API (POST JSON)
• Parsing of signals[] including multiple signal sets (e.g. boiler vs heating)
• Robust handling of 24:00 and cross-midnight low-tariff windows
• Per-signal schedule utilities:
  • current tariff (NT/VT)
  • current window start/end
  • next switch time
  • next NT/VT window (future-only)
  • remaining time until next switch
• High-level service (TariffService) that:
  • refreshes data occasionally (API call)
  • computes “snapshots” frequently without extra network calls (ideal for HA)

See examples/ for runnable demos (e.g., demo_cli.py).

Requirements

• Python >= 3.13
• Runtime dependency: httpx

Development tools (optional): uv, ruff, pyright, pytest, pytest-asyncio.

Version

The package version is derived from git tags via uv-dynamic-versioning.

from cez_distribution_hdo import __version__

print(__version__)

Install

From PyPI (recommended)

pip install cez-distribution-hdo

Using uv:

uv add cez-distribution-hdo

From TestPyPI (pre-releases)

TestPyPI is useful for verifying releases before publishing to PyPI. Pre-releases may require --pre.

With pip:

pip install --index-url https://test.pypi.org/simple/ --extra-index-url https://pypi.org/simple cez-distribution-hdo

With uv (using the testpypi index from pyproject.toml):

uv add --index testpypi cez-distribution-hdo

From Git (development)

Using uv:

uv add "cez-distribution-hdo @ git+https://github.com/pokornyIt/cez-distribution-hdo.git"

Or with pip:

pip install "cez-distribution-hdo @ git+https://github.com/pokornyIt/cez-distribution-hdo.git"

Logging

This library uses Python's standard logging module and does not configure logging by itself. To see debug logs from the library, configure logging in your application:

import logging

logging.basicConfig(level=logging.INFO)

# Increase verbosity for this library
logging.getLogger("cez_distribution_hdo").setLevel(logging.DEBUG)

If you also want to see HTTPX request logs:

import logging

logging.getLogger("httpx").setLevel(logging.INFO)

Quickstart

1) Fetch schedules (API call)

import asyncio

from cez_distribution_hdo import CezHdoClient


async def main() -> None:
    async with CezHdoClient() as client:
        # Provide exactly one identifier: ean OR sn OR place
        resp = await client.fetch_signals(ean="859182400123456789")
        print(f"Signals returned: {len(resp.data.signals)}")
        for s in resp.data.signals[:3]:
            print(s.signal, s.date_str, s.times_raw)


if __name__ == "__main__":
    asyncio.run(main())

2) High-level service (recommended)

Refresh schedules occasionally (e.g. hourly) and compute values frequently (e.g. every 1–5 seconds) without extra API calls.

import asyncio
from datetime import datetime
from zoneinfo import ZoneInfo

from cez_distribution_hdo import TariffService, snapshot_to_dict


async def main() -> None:
    tz = ZoneInfo("Europe/Prague")
    svc = TariffService(tz_name="Europe/Prague")

    # One API call (do this occasionally)
    # Provide exactly one identifier: ean OR sn OR place
    await svc.refresh(ean="859182400123456789")

    print("Available signals:", svc.signals)

    # Compute values (no network) - do this often
    now = datetime.now(tz)
    for signal in svc.signals:
        snap = svc.snapshot(signal, now=now)
        print(snapshot_to_dict(snap))


if __name__ == "__main__":
    asyncio.run(main())

Identifiers (EAN / SN / place)

The CEZ Distribution API accepts exactly one identifier per request.

Provide one of:

• ean — EAN of the electricity meter
• sn — serial number of the electricity meter
• place — place number of the electricity meter

If you pass none or more than one, the library raises InvalidRequestError.

Service exports (state & data)

After refresh(), the service keeps the latest data in memory and exposes it in three levels:

• Raw (original):

  • TariffService.last_response – last parsed response object (or None)
  • TariffService.last_response_raw() – raw API data dict (debug)

• Enriched (parsed schedules):

  • TariffService.schedules – read-only mapping {signal: SignalSchedule}
  • TariffService.get_schedule(signal) – one schedule by signal

• Curated (ready-to-use snapshots):

  • TariffService.snapshot(signal) – one computed snapshot
  • TariffService.snapshots_dict() – computed snapshots for all signals (dicts)

Example

import asyncio
from pprint import pprint
from cez_distribution_hdo import TariffService

async def main() -> None:
    svc = TariffService()
    await svc.refresh(ean="859182400123456789")

    print("Last refresh UTC:", svc.last_refresh_iso_utc)
    print("Signals:", svc.signals)

    # raw payload (debug)
    print("Raw keys:", list((svc.last_response_raw() or {}).keys()))

    # curated export
    pprint(svc.snapshots_dict())

if __name__ == "__main__":
    asyncio.run(main())

Data model

The API returns a list of signal entries:

• signal – identifies a “signal set” (multiple sets may be returned)
• datum – date (DD.MM.YYYY)
• casy – semicolon-separated time ranges where low tariff (NT) is active (everything outside those windows is high tariff (VT))

Example:

{
  "signal": "PTV2",
  "datum": "03.01.2026",
  "casy": "00:00-06:00; 17:00-24:00"
}

Cross-midnight handling

24:00 is treated as 00:00 of the next day.

If a low-tariff window ends at 24:00 and the next day starts with 00:00-06:00, the library merges these into one continuous interval:

• 03.01 17:00 → 04.01 06:00

This makes “current window”, “next switch”, and “remaining time” behave correctly.

Refresh behavior (midnight carry)

The CEZ Distribution API may drop “yesterday” shortly after midnight. That can break cross-midnight low-tariff windows (e.g. 17:00–24:00 + 00:00–06:00 should be one continuous interval).

To keep schedule computations stable, TariffService.refresh() automatically carries the previous day (D-1) from the last successful refresh when needed:

• For each signal present in the new response, if day D-1 is missing, it is taken from the previous refresh and merged in.

• If a signal is missing entirely in the new response, the service may carry:

  • D-1 and
  • all entries for D and later from the previous refresh,
  • but only if there is at least some data for D or later (otherwise the signal is dropped).

• Duplicates are removed (stable, deterministic de-duplication).

This means TariffService.last_response and computed schedules may represent an enriched view of API data across refreshes.

Error handling

The client raises:

• InvalidRequestError – invalid request (must provide exactly one identifier: ean/sn/place)
• HttpRequestError – network/timeout/non-2xx HTTP errors
• InvalidResponseError – unexpected JSON schema or invalid time/date formats
• ApiError – API returned non-200 statusCode in JSON payload

Development

Setup

uv venv
uv sync

Lint / typecheck / tests

uv run ruff check .
uv run pyright
uv run pytest

Pre-commit

uv add --group dev pre-commit
uv run pre-commit install
uv run pre-commit run --all-files

Build

uv build

License

This project is licensed under the MIT License. See LICENSE for details.