"""
P86 — Branchen-Benchmark.

Vergleicht den eigenen Compliance-Score mit dem Branchen-Median aus
allen bisherigen Snapshots derselben industry (P79 scan_context).
Liefert: "Sie 42% — Automotive-Median 58% (Stichprobe: 12 Sites)".

Wird in der Mail-Composition direkt unter dem Score im GF-1-Pager
gerendert. Mindest-Stichprobe = 3 vergleichbare Snapshots, sonst skip.

Heuristik fuer Score-Extraktion aus banner_result:
- banner_result.completeness_pct ODER
- banner_result.correctness_pct ODER
- 100 - len(banner_checks.violations) * 5 als Fallback.
"""

from __future__ import annotations

import json
import logging
from typing import Any

from sqlalchemy import text
from sqlalchemy.orm import Session

logger = logging.getLogger(__name__)

_MIN_SAMPLE = 3


def _extract_score(banner_result: dict | None) -> float | None:
    if not isinstance(banner_result, dict):
        return None
    for key in ("compliance_score", "completeness_pct", "correctness_pct"):
        v = banner_result.get(key)
        if isinstance(v, (int, float)):
            return float(v)
    bc = banner_result.get("banner_checks") or {}
    if isinstance(bc, dict):
        viols = bc.get("violations") or []
        if isinstance(viols, list):
            return max(0.0, 100.0 - len(viols) * 5)
    return None


def compute_benchmark(
    db: Session,
    industry: str,
    current_score: float | None,
    current_check_id: str,
) -> dict | None:
    if not industry or current_score is None:
        return None
    # Snapshots mit gleicher industry in scan_context.
    rows = db.execute(text(
        """
        SELECT banner_result FROM compliance.compliance_check_snapshots
        WHERE check_id != :cid
          AND scan_context IS NOT NULL
          AND scan_context->>'industry' = :ind
        ORDER BY created_at DESC
        LIMIT 50
        """
    ), {"cid": current_check_id, "ind": industry}).fetchall()
    scores: list[float] = []
    for r in rows:
        br = r[0]
        if isinstance(br, str):
            try:
                br = json.loads(br)
            except Exception:
                continue
        s = _extract_score(br)
        if s is not None:
            scores.append(s)
    if len(scores) < _MIN_SAMPLE:
        return None
    scores.sort()
    n = len(scores)
    median = scores[n // 2] if n % 2 else (scores[n // 2 - 1] + scores[n // 2]) / 2
    pct_lower = round(sum(1 for s in scores if s < current_score) / n * 100)
    return {
        "industry":  industry,
        "current":   round(current_score, 1),
        "median":    round(median, 1),
        "sample_size": n,
        "percentile": pct_lower,  # 80 = besser als 80% der Branche
    }


def build_benchmark_html(bench: dict) -> str:
    if not bench:
        return ""
    delta = bench["current"] - bench["median"]
    if delta >= 5:
        color = "#16a34a"
        verdict = "ueber dem Branchen-Median"
    elif delta <= -5:
        color = "#dc2626"
        verdict = "unter dem Branchen-Median"
    else:
        color = "#ca8a04"
        verdict = "etwa auf Branchen-Median"
    return (
        '<div style="font-family:-apple-system,BlinkMacSystemFont,sans-serif;'
        'max-width:760px;margin:0 auto 12px;padding:8px 14px;'
        'background:#f0f9ff;border:1px solid #bfdbfe;border-radius:6px;'
        'font-size:11px;color:#1e293b">'
        f'<strong>Branchen-Vergleich ({bench["industry"]}):</strong> '
        f'Ihr Score <strong>{bench["current"]:.1f}</strong> '
        f'<span style="color:{color}">({verdict}, '
        f'Median {bench["median"]:.1f})</span>. '
        f'<span style="color:#64748b">Sie sind besser als '
        f'{bench["percentile"]}% der bisher von uns gepruften '
        f'{bench["sample_size"]} Sites in dieser Branche.</span>'
        '</div>'
    )