ClimatController/app/metrics/service.py

import asyncio
import logging
import time
from pathlib import Path


logger = logging.getLogger(__name__)


class PrometheusMetricsService:

    def __init__(
        self,
        tion_service,
        qingping_service,
        schedule_service=None,
        auto_controller=None,
        output_path: str | Path = (
            "/var/lib/prometheus/node-exporter/"
            "climatcontroller.prom"
        ),
        interval: float = 5.0,
    ):
        self._tion = tion_service
        self._qingping = qingping_service
        self._schedule = schedule_service
        self._auto = auto_controller

        self._output_path = Path(output_path)
        self._interval = interval

        self._running = False
        self._task = None
        self._last_error: str | None = None

    @property
    def running(self) -> bool:
        return self._running

    @property
    def last_error(self) -> str | None:
        return self._last_error

    async def start(self) -> None:

        if self._running:
            return

        self._running = True

        await self._write_metrics()

        self._task = asyncio.create_task(
            self._loop()
        )

    async def stop(self) -> None:

        if not self._running:
            return

        self._running = False

        if self._task is not None:
            self._task.cancel()

            try:
                await self._task
            except asyncio.CancelledError:
                pass

            self._task = None

    async def _loop(self) -> None:

        while self._running:

            await asyncio.sleep(
                self._interval
            )

            try:
                await self._write_metrics()

            except Exception as exc:
                self._last_error = f"{type(exc).__name__}: {exc}"
                logger.exception("Prometheus metrics update failed")

    async def _write_metrics(self) -> None:

        lines: list[str] = []

        # --------------------------------------------------
        # Время последнего успешного обновления файла.
        # Позволяет определить протухшие метрики.
        # --------------------------------------------------

        self._add_gauge(
            lines,
            "climat_metrics_timestamp_seconds",
            time.time(),
            "Unix timestamp of the last "
            "ClimatController metrics update",
        )

        # --------------------------------------------------
        # Tion
        # --------------------------------------------------

        self._add_gauge(
            lines,
            "climat_tion_online",
            self._tion.online,
            "Tion connection state",
        )

        tion_state = self._tion.state

        if tion_state is not None:

            self._add_gauge(
                lines,
                "climat_tion_power",
                tion_state.power,
                "Tion power state",
            )

            self._add_gauge(
                lines,
                "climat_tion_heater",
                tion_state.heater,
                "Tion heater enabled state",
            )

            self._add_gauge(
                lines,
                "climat_tion_heating",
                tion_state.heating,
                "Tion active heating state",
            )

            self._add_gauge(
                lines,
                "climat_tion_fan_speed",
                tion_state.fan_speed,
                "Tion fan speed",
            )

            self._add_gauge(
                lines,
                "climat_tion_in_temperature_celsius",
                tion_state.in_temp,
                "Tion inlet temperature",
            )

            self._add_gauge(
                lines,
                "climat_tion_out_temperature_celsius",
                tion_state.out_temp,
                "Tion outside temperature",
            )

            self._add_gauge(
                lines,
                "climat_tion_target_temperature_celsius",
                tion_state.target_temp,
                "Tion target temperature",
            )

        # --------------------------------------------------
        # Qingping
        # --------------------------------------------------

        self._add_gauge(
            lines,
            "climat_qingping_online",
            self._qingping.online,
            "Qingping online state",
        )

        qingping_state = self._qingping.state

        if qingping_state is not None:

            self._add_gauge(
                lines,
                "climat_qingping_mqtt_connected",
                qingping_state.mqtt_connected,
                "Qingping MQTT connection state",
            )

            self._add_gauge(
                lines,
                "climat_qingping_temperature_celsius",
                qingping_state.temperature,
                "Qingping temperature",
            )

            self._add_gauge(
                lines,
                "climat_qingping_humidity_percent",
                qingping_state.humidity,
                "Qingping relative humidity",
            )

            self._add_gauge(
                lines,
                "climat_qingping_co2_ppm",
                qingping_state.co2,
                "Qingping CO2 concentration",
            )

            self._add_gauge(
                lines,
                "climat_qingping_pm25_ug_m3",
                qingping_state.pm25,
                "Qingping PM2.5 concentration",
            )

            self._add_gauge(
                lines,
                "climat_qingping_pm10_ug_m3",
                qingping_state.pm10,
                "Qingping PM10 concentration",
            )

            self._add_gauge(
                lines,
                "climat_qingping_battery_percent",
                qingping_state.battery,
                "Qingping battery level",
            )

        # --------------------------------------------------
        # Schedule / AUTO
        # --------------------------------------------------

        if self._schedule is not None:

            self._add_gauge(
                lines,
                "climat_schedule_paused",
                self._schedule.paused,
                "Schedule manual pause state",
            )

            try:
                resolution = (
                    self._schedule.resolve()
                )

                self._add_gauge(
                    lines,
                    "climat_auto_active",
                    resolution.auto_active,
                    "AUTO schedule state",
                )

            except Exception:
                logger.exception(
                    "Schedule metrics resolve failed"
                )

        # --------------------------------------------------
        # Atomic write
        # --------------------------------------------------

        self._output_path.parent.mkdir(
            parents=True,
            exist_ok=True,
        )

        temp_path = (
            self._output_path.with_suffix(
                self._output_path.suffix
                + ".tmp"
            )
        )

        temp_path.write_text(
            "\n".join(lines) + "\n",
            encoding="utf-8",
        )

        temp_path.replace(
            self._output_path
        )

        self._last_error = None

    @staticmethod
    def _add_gauge(
        lines: list[str],
        name: str,
        value,
        help_text: str,
    ) -> None:

        if value is None:
            return

        if isinstance(value, bool):
            value = 1 if value else 0

        lines.append(f"# HELP {name} {help_text}")

        lines.append(f"# TYPE {name} gauge")

        lines.append(f"{name} {value}")