"""
enrich_router.py — Round-2 enrichment endpoint
Author: dradulic@outlook.com  Date: 2026-05-04

Surfaces "Obogati podatke" buttons for klubovi, savezi, sportasi.

Strategy:
  1) Read what's already in DB and surface fields the frontend may not have shown.
  2) Build curated research URLs (Google, Wikipedia HR, Sportilus, sport-pgz.hr,
     HNS Semafor) so the operator can verify or expand by hand.
  3) If the entity has a `web` URL set, quickly fetch the page and extract
     <title> + <meta description> to return as a "live snippet". 5s timeout, fail-soft.
"""
import os, re, json, time, urllib.parse, urllib.request, html
import psycopg2, psycopg2.extras
from fastapi import APIRouter, HTTPException

router = APIRouter()

DB = dict(host=os.environ.get('PG_HOST','10.10.0.2'),
          port=int(os.environ.get('PG_PORT','6432')),
          dbname=os.environ.get('PG_DB','rinet_v3'),
          user=os.environ.get('PG_USER','rinet'),
          password=os.environ.get('PG_PASS',''))

UA = 'pgz-sport-enrich/2.0'

def _db():
    c = psycopg2.connect(**DB); c.autocommit = True; return c

def _fetch_one(sql, p):
    with _db() as c, c.cursor(cursor_factory=psycopg2.extras.RealDictCursor) as cur:
        cur.execute(sql, p)
        r = cur.fetchone()
        return dict(r) if r else None

def _fetch_title(url, timeout=5):
    if not url: return None
    try:
        if not url.startswith('http'):
            return None
        req = urllib.request.Request(url, headers={'User-Agent': UA})
        with urllib.request.urlopen(req, timeout=timeout) as r:
            data = r.read(40000).decode('utf-8','ignore')
        title_m = re.search(r'<title[^>]*>([^<]+)</title>', data, re.I)
        desc_m = re.search(r'<meta\s+name=["\']description["\']\s+content=["\']([^"\']+)["\']', data, re.I)
        og_desc_m = re.search(r'<meta\s+property=["\']og:description["\']\s+content=["\']([^"\']+)["\']', data, re.I)
        return {
            'url': url,
            'title': html.unescape(title_m.group(1).strip())[:300] if title_m else None,
            'description': html.unescape((desc_m or og_desc_m).group(1).strip())[:500] if (desc_m or og_desc_m) else None,
            'fetched_at': int(time.time()),
        }
    except Exception as e:
        return {'url': url, 'error': str(e)[:120]}

def _research_links(naziv, kind, grad=None):
    base_q = (naziv or '').strip()
    if grad: q = base_q + ' ' + grad
    else: q = base_q
    qenc = urllib.parse.quote(q)
    out = [
        {'label':'Google', 'icon':'🔍', 'url':'https://www.google.com/search?q='+qenc},
        {'label':'Wikipedia HR', 'icon':'📚', 'url':'https://hr.wikipedia.org/w/index.php?search='+qenc},
        {'label':'sport-pgz.hr', 'icon':'🏅', 'url':'https://sport-pgz.hr/?s='+qenc},
    ]
    if kind == 'klub':
        out.append({'label':'Sportilus', 'icon':'⬡', 'url':'https://www.sportilus.com/?s='+qenc})
        out.append({'label':'Sudski registar', 'icon':'⚖', 'url':'https://sudreg.pravosudje.hr/registar/oc/index.html'})
    if kind == 'sportas':
        out.append({'label':'HNS Semafor', 'icon':'⚽', 'url':'https://semafor.hns.family/?s='+qenc})
        out.append({'label':'transfermarkt', 'icon':'⚽', 'url':'https://www.transfermarkt.com/schnellsuche/ergebnis/schnellsuche?query='+qenc})
    if kind == 'savez':
        out.append({'label':'sport-pgz.hr savezi', 'icon':'🏅', 'url':'https://sport-pgz.hr/savezi'})
    return out

@router.post("/enrich/{kind}/{eid}")
def enrich(kind: str, eid: int):
    if kind not in ('klub','savez','sportas'):
        raise HTTPException(400, "kind must be klub|savez|sportas")

    if kind == 'klub':
        row = _fetch_one("""SELECT id, naziv, oib, sport, grad, predsjednik, tajnik,
                                   web, web_stranica, email, telefon, ciljevi, opis_djelatnosti,
                                   sjediste, godina_osnutka, savez_id, scrape_url, source_url
                            FROM pgz_sport.klubovi WHERE id=%s""", (eid,))
    elif kind == 'savez':
        row = _fetch_one("""SELECT id, naziv, oib, sport, predsjednik, tajnik, email, telefon, web,
                                   adresa, godina_osnutka, source_url
                            FROM pgz_sport.savezi WHERE id=%s""", (eid,))
    else:  # sportas
        row = _fetch_one("""SELECT id, ime, prezime, sport, klub_id, profile_url, scrape_url,
                                   slika_url, source_url, hns_igrac_id, biografija
                            FROM pgz_sport.clanovi WHERE id=%s""", (eid,))
    if not row:
        raise HTTPException(404, kind+" not found")

    # Build display name
    if kind == 'sportas':
        naziv = (row.get('ime','') + ' ' + row.get('prezime','')).strip()
        grad = None
    else:
        naziv = row.get('naziv','')
        grad = row.get('grad') if kind=='klub' else None

    # Live web snippet from primary URL
    primary = row.get('web') or row.get('web_stranica') or row.get('source_url') or row.get('scrape_url') or row.get('profile_url')
    snippet = _fetch_title(primary) if primary else None

    # Coverage score: how many key fields are filled?
    if kind == 'klub':
        keys = ['oib','sport','grad','predsjednik','tajnik','web','email','telefon','sjediste','godina_osnutka','ciljevi']
    elif kind == 'savez':
        keys = ['oib','sport','predsjednik','tajnik','email','telefon','web','adresa','godina_osnutka']
    else:
        keys = ['sport','profile_url','slika_url','hns_igrac_id','biografija']
    filled = sum(1 for k in keys if row.get(k))
    coverage = round(filled/len(keys)*100)

    # Suggested missing fields
    missing = [k for k in keys if not row.get(k)]

    return {
        'kind': kind,
        'id': eid,
        'naziv': naziv,
        'coverage': coverage,
        'filled_fields': filled,
        'total_fields': len(keys),
        'missing_fields': missing,
        'live_snippet': snippet,
        'research_links': _research_links(naziv, kind, grad),
        'enriched_at': int(time.time()),
    }