#!/usr/bin/env python3
"""
coverage_report.py — Per-entity coverage scoring across pgz_sport schema

Fills /opt/pgz-sport/data_quality_report.md with:
  - per-type aggregate (n, mean coverage, median, # zero-coverage, # complete)
  - distribution histogram
  - top 50 entities most needing manual review (lowest coverage AND non-empty name)
  - link to detail panel for each (so audit.html-style triage is one click away)
"""
import os, json
from collections import Counter
from datetime import datetime, timezone
import psycopg2, psycopg2.extras

PG = dict(host='10.10.0.2', port=6432, dbname='rinet_v3',
          user='rinet', password=os.environ["DB_PASSWORD"])

# Per-type coverage definition: list of fields that count toward coverage
DEFS = {
    'savez': {
        'table': 'pgz_sport.savezi',
        'name_col': 'naziv',
        'fields': ['naziv','sport','predsjednik','tajnik','email','telefon','web','oib','adresa','godina_osnutka'],
        'panel_path': lambda i: f'/?nav=savezi&open={i}',
    },
    'klub': {
        'table': 'pgz_sport.klubovi',
        'name_col': 'naziv',
        # Use COALESCE-ish: web OR web_stranica counts; sjediste OR adresa counts
        'fields': ['naziv','sport','grad','oib','predsjednik','tajnik','email','telefon',
                   'web_or_stranica','sjediste_or_adresa','ciljevi','opis_djelatnosti'],
        'panel_path': lambda i: f'/?nav=klubovi&open={i}',
    },
    'sportas': {
        'table': 'pgz_sport.clanovi',
        'name_col': "ime||' '||prezime",
        'fields': ['ime','prezime','sport','klub_id','datum_rodenja','slika_url','oib','profile_url','biografija','hns_igrac_id'],
        'panel_path': lambda i: f'/?nav=sportasi&open={i}',
    },
    'objekt': {
        'table': 'pgz_sport.sportski_objekti',
        'name_col': 'naziv',
        'fields': ['naziv','tip','grad','adresa','lat','lng','upravitelj','kapacitet','sportovi','izgradeno'],
        'panel_path': lambda i: f'/?nav=objekti&open={i}',
    },
    'manifestacija': {
        'table': 'pgz_sport.manifestacije',
        'name_col': 'naziv',
        'fields': ['naziv','mjesto','organizator','razina','broj_ucesnika','godina_od','source_url'],
        'panel_path': lambda i: f'/?nav=manifestacije&open={i}',
    },
}

def fetch_rows(cur, kind: str):
    spec = DEFS[kind]
    table = spec['table']
    if kind == 'klub':
        sql = f"""
          SELECT id, naziv,
            (CASE WHEN naziv IS NOT NULL AND naziv<>'' THEN 1 ELSE 0 END +
             CASE WHEN sport IS NOT NULL AND sport<>'' THEN 1 ELSE 0 END +
             CASE WHEN grad IS NOT NULL AND grad<>'' THEN 1 ELSE 0 END +
             CASE WHEN oib IS NOT NULL AND oib<>'' THEN 1 ELSE 0 END +
             CASE WHEN predsjednik IS NOT NULL AND predsjednik<>'' THEN 1 ELSE 0 END +
             CASE WHEN tajnik IS NOT NULL AND tajnik<>'' THEN 1 ELSE 0 END +
             CASE WHEN email IS NOT NULL AND email<>'' THEN 1 ELSE 0 END +
             CASE WHEN telefon IS NOT NULL AND telefon<>'' THEN 1 ELSE 0 END +
             CASE WHEN COALESCE(web, web_stranica) IS NOT NULL AND COALESCE(web, web_stranica)<>'' THEN 1 ELSE 0 END +
             CASE WHEN COALESCE(sjediste, adresa) IS NOT NULL AND COALESCE(sjediste, adresa)<>'' THEN 1 ELSE 0 END +
             CASE WHEN ciljevi IS NOT NULL AND ciljevi<>'' THEN 1 ELSE 0 END +
             CASE WHEN opis_djelatnosti IS NOT NULL AND opis_djelatnosti<>'' THEN 1 ELSE 0 END
            ) AS filled
          FROM {table}
        """
    elif kind == 'sportas':
        sql = f"""
          SELECT id, (COALESCE(ime,'')||' '||COALESCE(prezime,'')) AS naziv,
            (CASE WHEN ime IS NOT NULL AND ime<>'' THEN 1 ELSE 0 END +
             CASE WHEN prezime IS NOT NULL AND prezime<>'' THEN 1 ELSE 0 END +
             CASE WHEN sport IS NOT NULL AND sport<>'' THEN 1 ELSE 0 END +
             CASE WHEN klub_id IS NOT NULL THEN 1 ELSE 0 END +
             CASE WHEN datum_rodenja IS NOT NULL THEN 1 ELSE 0 END +
             CASE WHEN slika_url IS NOT NULL AND slika_url<>'' THEN 1 ELSE 0 END +
             CASE WHEN oib IS NOT NULL AND oib<>'' THEN 1 ELSE 0 END +
             CASE WHEN profile_url IS NOT NULL AND profile_url<>'' THEN 1 ELSE 0 END +
             CASE WHEN biografija IS NOT NULL AND biografija<>'' THEN 1 ELSE 0 END +
             CASE WHEN hns_igrac_id IS NOT NULL AND hns_igrac_id<>'' THEN 1 ELSE 0 END
            ) AS filled
          FROM {table}
        """
    elif kind == 'objekt':
        sql = f"""
          SELECT id, naziv,
            (CASE WHEN naziv IS NOT NULL AND naziv<>'' THEN 1 ELSE 0 END +
             CASE WHEN tip IS NOT NULL AND tip<>'' THEN 1 ELSE 0 END +
             CASE WHEN grad IS NOT NULL AND grad<>'' THEN 1 ELSE 0 END +
             CASE WHEN adresa IS NOT NULL AND adresa<>'' THEN 1 ELSE 0 END +
             CASE WHEN lat IS NOT NULL THEN 1 ELSE 0 END +
             CASE WHEN lng IS NOT NULL THEN 1 ELSE 0 END +
             CASE WHEN upravitelj IS NOT NULL AND upravitelj<>'' THEN 1 ELSE 0 END +
             CASE WHEN kapacitet IS NOT NULL THEN 1 ELSE 0 END +
             CASE WHEN sportovi IS NOT NULL AND array_length(sportovi,1)>0 THEN 1 ELSE 0 END +
             CASE WHEN izgradeno IS NOT NULL THEN 1 ELSE 0 END
            ) AS filled
          FROM {table}
        """
    elif kind == 'manifestacija':
        sql = f"""
          SELECT id, naziv,
            (CASE WHEN naziv IS NOT NULL AND naziv<>'' THEN 1 ELSE 0 END +
             CASE WHEN mjesto IS NOT NULL AND mjesto<>'' THEN 1 ELSE 0 END +
             CASE WHEN organizator IS NOT NULL AND organizator<>'' THEN 1 ELSE 0 END +
             CASE WHEN razina IS NOT NULL AND razina<>'' THEN 1 ELSE 0 END +
             CASE WHEN broj_ucesnika IS NOT NULL AND broj_ucesnika::text<>'' THEN 1 ELSE 0 END +
             CASE WHEN godina_od IS NOT NULL THEN 1 ELSE 0 END +
             CASE WHEN source_url IS NOT NULL AND source_url<>'' THEN 1 ELSE 0 END
            ) AS filled
          FROM {table}
        """
    else:  # savez
        sql = f"""
          SELECT id, naziv,
            (CASE WHEN naziv IS NOT NULL AND naziv<>'' THEN 1 ELSE 0 END +
             CASE WHEN sport IS NOT NULL AND sport<>'' THEN 1 ELSE 0 END +
             CASE WHEN predsjednik IS NOT NULL AND predsjednik<>'' THEN 1 ELSE 0 END +
             CASE WHEN tajnik IS NOT NULL AND tajnik<>'' THEN 1 ELSE 0 END +
             CASE WHEN email IS NOT NULL AND email<>'' THEN 1 ELSE 0 END +
             CASE WHEN telefon IS NOT NULL AND telefon<>'' THEN 1 ELSE 0 END +
             CASE WHEN web IS NOT NULL AND web<>'' THEN 1 ELSE 0 END +
             CASE WHEN oib IS NOT NULL AND oib<>'' THEN 1 ELSE 0 END +
             CASE WHEN adresa IS NOT NULL AND adresa<>'' THEN 1 ELSE 0 END +
             CASE WHEN godina_osnutka IS NOT NULL THEN 1 ELSE 0 END
            ) AS filled
          FROM {table}
        """
    cur.execute(sql)
    rows = []
    for r in cur.fetchall():
        rows.append({'kind': kind, 'id': r['id'], 'naziv': r['naziv'] or '',
                     'filled': int(r['filled']),
                     'total': len(spec['fields'])})
    return rows


def stats(rows):
    if not rows: return {}
    pcts = [r['filled']/r['total']*100 for r in rows]
    pcts.sort()
    n = len(pcts)
    mean = sum(pcts)/n
    median = pcts[n//2]
    zero = sum(1 for p in pcts if p == 0)
    complete = sum(1 for p in pcts if p >= 99.0)
    bins = Counter()
    for p in pcts:
        b = int(p // 10) * 10
        if b == 100: b = 90
        bins[b] += 1
    return {'n': n, 'mean': round(mean,1), 'median': round(median,1),
            'zero': zero, 'complete': complete,
            'distribution': dict(sorted(bins.items()))}


def main():
    conn = psycopg2.connect(**PG)
    cur = conn.cursor(cursor_factory=psycopg2.extras.RealDictCursor)
    all_rows = []
    by_kind = {}
    for kind in DEFS:
        rows = fetch_rows(cur, kind)
        by_kind[kind] = rows
        all_rows.extend(rows)
        print(f'{kind:14s} n={len(rows):5d} mean={stats(rows)["mean"]:.1f}% complete={stats(rows)["complete"]}')

    # Top 50 worst — exclude rows with empty naziv (those are flagged separately)
    valid = [r for r in all_rows if (r['naziv'] or '').strip()]
    # Sort by coverage ASC, then by total DESC
    worst = sorted(valid, key=lambda r: (r['filled']/r['total'], -r['total']))[:50]
    out = {
        'generated_at': datetime.now(timezone.utc).isoformat(),
        'totals': {k: len(v) for k,v in by_kind.items()},
        'total_entities': len(all_rows),
        'per_type_stats': {k: stats(v) for k,v in by_kind.items()},
        'top50_review': worst,
    }
    print(f'\nTotal entities: {len(all_rows)}')
    print(f'Top 50 worst — sample:')
    for r in worst[:5]:
        pct = r['filled']/r['total']*100
        print(f"  {r['kind']:14s} id={r['id']:7d} {r['naziv'][:50]:50s} {r['filled']}/{r['total']} ({pct:.0f}%)")
    json.dump(out, open('/tmp/coverage_data.json','w'), ensure_ascii=False, default=str)
    cur.close(); conn.close()


if __name__ == '__main__':
    main()