intelligence/username_intel.py

from __future__ import annotations

import math
import re
from collections import Counter
from typing import List

from shield.models.evidence import Evidence


def _entropy(value: str) -> float:
    counts = Counter(value)
    total = len(value)
    return -sum((count / total) * math.log2(count / total) for count in counts.values())


def collect_username_evidence(username: str | None) -> List[Evidence]:
    if not username:
        return []
    evidence: List[Evidence] = []
    if len(username) >= 10 and _entropy(username) >= 3.2 and re.search(r"\d{4,}$", username):
        evidence.append(Evidence(
            source="username_intelligence",
            category="synthetic_identity",
            risk_score=42,
            weight=0.45,
            confidence=0.72,
            explanation="Username resembles generated account-farm naming",
        ))
    if re.fullmatch(r"[a-z]{1,4}\d{6,}", username):
        evidence.append(Evidence(
            source="username_intelligence",
            category="synthetic_identity",
            risk_score=48,
            weight=0.5,
            confidence=0.78,
            explanation="Username has a compact random-prefix numeric pattern",
        ))
    return evidence