Trixy7/trixy_core/satellite/metrics/aggregator.py

# -*- coding: utf-8 -*-
"""
Metriken-Aggregator für Satellites.

Aggregiert Metriken über Zeit und Satellites
für Analyse und Reporting.
"""

from __future__ import annotations

import logging
import statistics
from collections import defaultdict
from dataclasses import dataclass, field
from datetime import datetime, timedelta
from enum import Enum, auto
from typing import Any, Callable

from trixy_core.satellite.metrics.collector import (
    SatelliteMetricsCollector,
    MetricValue,
    SatelliteMetrics,
)


class AggregationType(Enum):
    """Aggregations-Typ."""
    SUM = auto()          # Summe
    AVG = auto()          # Durchschnitt
    MIN = auto()          # Minimum
    MAX = auto()          # Maximum
    COUNT = auto()        # Anzahl
    MEDIAN = auto()       # Median
    P95 = auto()          # 95. Perzentil
    P99 = auto()          # 99. Perzentil
    STDDEV = auto()       # Standardabweichung


class TimeWindow(Enum):
    """Zeitfenster für Aggregation."""
    MINUTE = 60
    FIVE_MINUTES = 300
    FIFTEEN_MINUTES = 900
    HOUR = 3600
    DAY = 86400


@dataclass
class AggregatedMetric:
    """Aggregierte Metrik."""
    name: str
    aggregation: AggregationType
    value: float
    window: TimeWindow
    sample_count: int
    start_time: datetime
    end_time: datetime
    satellite_ids: list[str] = field(default_factory=list)

    def to_dict(self) -> dict[str, Any]:
        """Konvertiert zu Dictionary."""
        return {
            "name": self.name,
            "aggregation": self.aggregation.name,
            "value": round(self.value, 4),
            "window": self.window.name,
            "sample_count": self.sample_count,
            "start_time": self.start_time.isoformat(),
            "end_time": self.end_time.isoformat(),
            "satellite_count": len(self.satellite_ids),
        }


class MetricsAggregator:
    """
    Aggregiert Metriken über Zeit und Satellites.

    Bietet verschiedene Aggregationsfunktionen und
    zeitbasierte Analysen.

    Beispiel:
        aggregator = MetricsAggregator(collector)

        # Durchschnittliche Latenz über alle Satellites
        avg_latency = aggregator.aggregate(
            metric_name="latency",
            aggregation=AggregationType.AVG,
            window=TimeWindow.FIVE_MINUTES,
        )

        # Top Satellites nach Fehlerrate
        top_errors = aggregator.get_top_satellites(
            metric_name="error_count",
            limit=10,
        )
    """

    def __init__(
        self,
        collector: SatelliteMetricsCollector,
        logger: logging.Logger | None = None,
    ) -> None:
        """
        Initialisiert den Aggregator.

        Args:
            collector: Metriken-Collector
            logger: Logger-Instanz
        """
        self.collector = collector
        self.logger = logger or logging.getLogger(__name__)

        # Cache für berechnete Aggregationen
        self._cache: dict[str, tuple[datetime, AggregatedMetric]] = {}
        self._cache_ttl = timedelta(seconds=30)

    def _get_metric_values(
        self,
        metric_name: str,
        window: TimeWindow,
        satellite_ids: list[str] | None = None,
    ) -> tuple[list[float], list[str]]:
        """
        Sammelt Metrik-Werte aus dem Collector.

        Args:
            metric_name: Name der Metrik
            window: Zeitfenster
            satellite_ids: Optional: Spezifische Satellites

        Returns:
            Tuple (Werte, Satellite-IDs)
        """
        cutoff = datetime.now() - timedelta(seconds=window.value)
        values: list[float] = []
        satellites: list[str] = []

        all_metrics = self.collector.get_all_metrics()

        for sat_id, metrics in all_metrics.items():
            if satellite_ids and sat_id not in satellite_ids:
                continue

            if metrics.last_updated < cutoff:
                continue

            # Standard-Metriken
            value = None
            if metric_name == "latency":
                value = metrics.latency_ms
            elif metric_name == "packet_loss":
                value = metrics.packet_loss_percent
            elif metric_name == "bandwidth":
                value = metrics.bandwidth_kbps
            elif metric_name == "messages_sent":
                value = float(metrics.messages_sent)
            elif metric_name == "messages_received":
                value = float(metrics.messages_received)
            elif metric_name == "error_count":
                value = float(metrics.error_count)
            elif metric_name == "reconnect_count":
                value = float(metrics.reconnect_count)

            if value is not None:
                values.append(value)
                satellites.append(sat_id)

        return values, satellites

    def _calculate_aggregation(
        self,
        values: list[float],
        aggregation: AggregationType,
    ) -> float:
        """Berechnet Aggregation für Werte."""
        if not values:
            return 0.0

        if aggregation == AggregationType.SUM:
            return sum(values)

        elif aggregation == AggregationType.AVG:
            return statistics.mean(values)

        elif aggregation == AggregationType.MIN:
            return min(values)

        elif aggregation == AggregationType.MAX:
            return max(values)

        elif aggregation == AggregationType.COUNT:
            return float(len(values))

        elif aggregation == AggregationType.MEDIAN:
            return statistics.median(values)

        elif aggregation == AggregationType.P95:
            return self._percentile(values, 95)

        elif aggregation == AggregationType.P99:
            return self._percentile(values, 99)

        elif aggregation == AggregationType.STDDEV:
            return statistics.stdev(values) if len(values) > 1 else 0.0

        return 0.0

    def _percentile(self, values: list[float], percentile: int) -> float:
        """Berechnet Perzentil."""
        if not values:
            return 0.0

        sorted_values = sorted(values)
        index = (percentile / 100) * (len(sorted_values) - 1)

        if index.is_integer():
            return sorted_values[int(index)]

        lower = sorted_values[int(index)]
        upper = sorted_values[int(index) + 1]
        fraction = index - int(index)

        return lower + (upper - lower) * fraction

    def aggregate(
        self,
        metric_name: str,
        aggregation: AggregationType = AggregationType.AVG,
        window: TimeWindow = TimeWindow.FIVE_MINUTES,
        satellite_ids: list[str] | None = None,
        use_cache: bool = True,
    ) -> AggregatedMetric:
        """
        Aggregiert eine Metrik.

        Args:
            metric_name: Name der Metrik
            aggregation: Aggregations-Typ
            window: Zeitfenster
            satellite_ids: Optional: Spezifische Satellites
            use_cache: Cache verwenden

        Returns:
            Aggregierte Metrik
        """
        # Cache prüfen
        cache_key = f"{metric_name}:{aggregation.name}:{window.name}"
        if use_cache and cache_key in self._cache:
            cached_time, cached_result = self._cache[cache_key]
            if datetime.now() - cached_time < self._cache_ttl:
                return cached_result

        # Werte sammeln
        values, satellites = self._get_metric_values(
            metric_name, window, satellite_ids
        )

        # Aggregieren
        result_value = self._calculate_aggregation(values, aggregation)

        end_time = datetime.now()
        start_time = end_time - timedelta(seconds=window.value)

        result = AggregatedMetric(
            name=metric_name,
            aggregation=aggregation,
            value=result_value,
            window=window,
            sample_count=len(values),
            start_time=start_time,
            end_time=end_time,
            satellite_ids=satellites,
        )

        # Cache aktualisieren
        self._cache[cache_key] = (datetime.now(), result)

        return result

    def aggregate_multiple(
        self,
        metric_names: list[str],
        aggregations: list[AggregationType],
        window: TimeWindow = TimeWindow.FIVE_MINUTES,
    ) -> dict[str, dict[str, AggregatedMetric]]:
        """
        Aggregiert mehrere Metriken mit mehreren Aggregationen.

        Args:
            metric_names: Liste von Metrik-Namen
            aggregations: Liste von Aggregations-Typen
            window: Zeitfenster

        Returns:
            Dict[metric_name][aggregation_name] -> AggregatedMetric
        """
        results: dict[str, dict[str, AggregatedMetric]] = {}

        for metric_name in metric_names:
            results[metric_name] = {}
            for aggregation in aggregations:
                result = self.aggregate(metric_name, aggregation, window)
                results[metric_name][aggregation.name] = result

        return results

    def get_top_satellites(
        self,
        metric_name: str,
        limit: int = 10,
        ascending: bool = False,
    ) -> list[tuple[str, float]]:
        """
        Holt Top-Satellites nach Metrik.

        Args:
            metric_name: Metrik-Name
            limit: Maximale Anzahl
            ascending: Aufsteigend sortieren

        Returns:
            Liste von (satellite_id, value) Tupeln
        """
        values, satellites = self._get_metric_values(
            metric_name,
            TimeWindow.HOUR,
        )

        if not values:
            return []

        # Zusammenführen
        satellite_values = list(zip(satellites, values))

        # Sortieren
        satellite_values.sort(key=lambda x: x[1], reverse=not ascending)

        return satellite_values[:limit]

    def get_metric_distribution(
        self,
        metric_name: str,
        window: TimeWindow = TimeWindow.HOUR,
        buckets: int = 10,
    ) -> dict[str, Any]:
        """
        Berechnet Verteilung einer Metrik.

        Args:
            metric_name: Metrik-Name
            window: Zeitfenster
            buckets: Anzahl der Buckets

        Returns:
            Verteilungs-Informationen
        """
        values, _ = self._get_metric_values(metric_name, window)

        if not values:
            return {
                "buckets": [],
                "min": 0,
                "max": 0,
                "count": 0,
            }

        min_val = min(values)
        max_val = max(values)

        if min_val == max_val:
            return {
                "buckets": [{"start": min_val, "end": max_val, "count": len(values)}],
                "min": min_val,
                "max": max_val,
                "count": len(values),
            }

        bucket_size = (max_val - min_val) / buckets
        bucket_counts = [0] * buckets

        for value in values:
            bucket_index = min(
                int((value - min_val) / bucket_size),
                buckets - 1
            )
            bucket_counts[bucket_index] += 1

        bucket_info = []
        for i in range(buckets):
            start = min_val + i * bucket_size
            end = start + bucket_size
            bucket_info.append({
                "start": round(start, 4),
                "end": round(end, 4),
                "count": bucket_counts[i],
            })

        return {
            "buckets": bucket_info,
            "min": min_val,
            "max": max_val,
            "count": len(values),
        }

    def get_satellite_comparison(
        self,
        metric_name: str,
        satellite_ids: list[str],
        window: TimeWindow = TimeWindow.HOUR,
    ) -> dict[str, dict[str, float]]:
        """
        Vergleicht Metrik über mehrere Satellites.

        Args:
            metric_name: Metrik-Name
            satellite_ids: Zu vergleichende Satellites
            window: Zeitfenster

        Returns:
            Dict mit Statistiken pro Satellite
        """
        result: dict[str, dict[str, float]] = {}

        for sat_id in satellite_ids:
            values, _ = self._get_metric_values(
                metric_name, window, [sat_id]
            )

            if values:
                result[sat_id] = {
                    "current": values[0] if values else 0.0,
                    "avg": statistics.mean(values) if values else 0.0,
                    "min": min(values) if values else 0.0,
                    "max": max(values) if values else 0.0,
                }
            else:
                result[sat_id] = {
                    "current": 0.0,
                    "avg": 0.0,
                    "min": 0.0,
                    "max": 0.0,
                }

        return result

    def get_health_summary(self) -> dict[str, Any]:
        """
        Erstellt Gesundheits-Übersicht basierend auf Metriken.

        Returns:
            Gesundheits-Übersicht
        """
        window = TimeWindow.FIVE_MINUTES

        avg_latency = self.aggregate("latency", AggregationType.AVG, window)
        p99_latency = self.aggregate("latency", AggregationType.P99, window)
        max_packet_loss = self.aggregate("packet_loss", AggregationType.MAX, window)
        total_errors = self.aggregate("error_count", AggregationType.SUM, window)

        # Gesundheitsstatus bestimmen
        health_status = "healthy"
        issues = []

        if avg_latency.value > 100:
            health_status = "degraded"
            issues.append(f"Hohe durchschnittliche Latenz: {avg_latency.value:.1f}ms")

        if p99_latency.value > 500:
            health_status = "degraded"
            issues.append(f"Hohe P99-Latenz: {p99_latency.value:.1f}ms")

        if max_packet_loss.value > 5:
            health_status = "degraded"
            issues.append(f"Hoher Paketverlust: {max_packet_loss.value:.1f}%")

        if total_errors.value > 10:
            health_status = "unhealthy"
            issues.append(f"Viele Fehler: {int(total_errors.value)}")

        return {
            "status": health_status,
            "issues": issues,
            "metrics": {
                "avg_latency_ms": round(avg_latency.value, 2),
                "p99_latency_ms": round(p99_latency.value, 2),
                "max_packet_loss_percent": round(max_packet_loss.value, 2),
                "total_errors": int(total_errors.value),
            },
            "satellite_count": avg_latency.sample_count,
        }

    def clear_cache(self) -> None:
        """Löscht den Cache."""
        self._cache.clear()