evidence/aggregator.py

from __future__ import annotations

from typing import Dict, Iterable, List

from shield.models.evidence import Evidence


def aggregate_evidence(evidence: Iterable[Evidence]) -> Dict[str, object]:
    items: List[Evidence] = list(evidence)
    if not items:
        return {
            "risk_score": 0,
            "confidence": 0.0,
            "reasons": [],
            "evidence": [],
            "threat_categories": [],
        }

    weighted_total = 0.0
    total_weight = 0.0
    max_risk = 0
    confidence_total = 0.0
    confidence_weight = 0.0
    reasons: List[str] = []
    categories: List[str] = []

    for item in items:
        influence = max(0.05, item.weight) * max(0.05, item.confidence)
        weighted_total += item.risk_score * influence
        total_weight += influence
        confidence_total += item.confidence * item.weight
        confidence_weight += item.weight
        max_risk = max(max_risk, item.risk_score)
        if item.explanation and item.explanation not in reasons:
            reasons.append(item.explanation)
        if item.category and item.category not in categories:
            categories.append(item.category)

    average_risk = weighted_total / total_weight if total_weight else 0.0
    risk_score = int(round((max_risk * 0.6) + (average_risk * 0.4)))
    confidence = confidence_total / confidence_weight if confidence_weight else 0.0

    return {
        "risk_score": max(0, min(100, risk_score)),
        "confidence": round(max(0.0, min(1.0, confidence)), 2),
        "reasons": reasons[:12],
        "evidence": [item.to_dict() for item in items],
        "threat_categories": categories,
    }