gendesign/site-finder/04_report.py

"""Generate JSON + HTML comparison report."""
import sqlite3, pathlib, json, statistics

DB = pathlib.Path(__file__).parent / "analysis.db"
OUT = pathlib.Path(__file__).parent / "reports"
OUT.mkdir(exist_ok=True)

PARCEL_ID = "parcel:66:41:0204016:10"

def fetch_economics(conn, site_id):
    row = conn.execute("""SELECT sd.district, sd.method, sd.nearest_jk_dist_m,
                                  de.n_projects, de.weighted_price_m2, de.median_price_m2,
                                  de.deals_per_month_avg, de.months_to_sellout,
                                  de.real_n_lots, de.real_n_sold, de.real_sold_pct,
                                  de.real_median_price_m2, de.real_p25_price_m2, de.real_p75_price_m2,
                                  de.real_avg_area_sold, de.real_velocity_per_month,
                                  de.real_avg_readiness_pct
                            FROM site_district sd
                            LEFT JOIN district_economics de USING (district)
                           WHERE sd.site_id=?""", (site_id,)).fetchone()
    if not row: return None
    cols = ["district","district_method","district_dist_m",
            "n_projects","weighted_price_m2_corp_sum","median_price_m2_corp_sum",
            "deals_per_month_corp_sum","months_to_sellout",
            "n_lots","n_sold","sold_pct",
            "median_price_m2","p25_price_m2","p75_price_m2",
            "avg_area_sold_m2","velocity_per_month","avg_readiness_pct"]
    return dict(zip(cols, row))


def fetch_site_full(conn, site_id):
    s = dict(zip(
        [c[0] for c in conn.execute("PRAGMA table_info(sites)").fetchall()],
        conn.execute("SELECT * FROM sites WHERE site_id=?", (site_id,)).fetchone()
    )) if False else None
    cur = conn.execute("SELECT * FROM sites WHERE site_id=?", (site_id,))
    cols = [c[0] for c in cur.description]
    s = dict(zip(cols, cur.fetchone()))
    s["features"] = dict(conn.execute(
        "SELECT feature,value FROM features WHERE site_id=?", (site_id,)).fetchall())
    s["scores"] = dict(conn.execute(
        "SELECT component,score_0_100 FROM scores WHERE site_id=?", (site_id,)).fetchall())
    tot = conn.execute(
        "SELECT weighted,rank_overall,rank_district FROM scores_total WHERE site_id=?",
        (site_id,)).fetchone()
    s["weighted_total"] = tot[0]
    s["rank_overall"] = tot[1]
    s["rank_district"] = tot[2]
    s["economics"] = fetch_economics(conn, site_id)
    s["nearest_pois"] = {}
    for cat in ["kindergarten","school","university","pharmacy","clinic","hospital",
                "shop_big","shop_med","shop_small","bus_stop","tram_stop","metro",
                "park","playground","sports"]:
        row = conn.execute(
            """SELECT name, distance_m, lat, lon FROM pois
               WHERE site_id=? AND category=? ORDER BY distance_m LIMIT 5""",
            (site_id, cat)).fetchall()
        s["nearest_pois"][cat] = [
            {"name": r[0] or "—", "distance_m": round(r[1], 1), "lat": r[2], "lon": r[3]}
            for r in row
        ]
    return s

def main():
    conn = sqlite3.connect(DB)
    parcel = fetch_site_full(conn, PARCEL_ID)
    n_total = conn.execute("SELECT count(*) FROM sites").fetchone()[0]

    # Top-10 best-scoring ЖК + parcel position context
    rows = conn.execute("""SELECT s.site_id, s.name, s.district, s.developer, s.obj_class,
                                  s.flat_count, s.lat, s.lon,
                                  st.weighted, st.rank_overall
                           FROM sites s JOIN scores_total st USING (site_id)
                           WHERE s.kind='jk' ORDER BY st.weighted DESC""").fetchall()
    top10 = [dict(zip(['site_id','name','district','developer','obj_class','flat_count',
                       'lat','lon','weighted','rank'], r)) for r in rows[:10]]

    # Distribution stats
    all_w = [r[8] for r in rows]
    stats = {
        "n_jk": len(all_w),
        "mean": round(statistics.mean(all_w), 1),
        "median": round(statistics.median(all_w), 1),
        "p25": round(statistics.quantiles(all_w, n=4)[0], 1),
        "p75": round(statistics.quantiles(all_w, n=4)[2], 1),
        "min": round(min(all_w), 1),
        "max": round(max(all_w), 1),
    }

    # Component-wise comparison
    parcel_comp = parcel["scores"]
    comp_stats = {}
    for c in parcel_comp:
        vals = [v for (v,) in conn.execute(
            "SELECT score_0_100 FROM scores s JOIN sites si USING (site_id) WHERE component=? AND si.kind='jk'",
            (c,)).fetchall()]
        comp_stats[c] = {
            "parcel": round(parcel_comp[c], 1),
            "median_jk": round(statistics.median(vals), 1),
            "p75_jk": round(statistics.quantiles(vals, n=4)[2], 1),
        }

    # Closest comparable ЖК (by geographic proximity to parcel)
    plat, plon = parcel["lat"], parcel["lon"]
    nearby = []
    for r in conn.execute("""SELECT s.site_id, s.name, s.district, s.developer, s.obj_class,
                                    s.lat, s.lon, st.weighted, st.rank_overall
                             FROM sites s JOIN scores_total st USING (site_id)
                             WHERE s.kind='jk'""").fetchall():
        import math
        R=6371000; p1,p2=math.radians(plat),math.radians(r[5])
        dp=math.radians(r[5]-plat); dl=math.radians(r[6]-plon)
        a=math.sin(dp/2)**2+math.cos(p1)*math.cos(p2)*math.sin(dl/2)**2
        d=2*R*math.asin(math.sqrt(a))
        nearby.append((d, r))
    nearby.sort()
    closest = [
        {"distance_m": round(d, 0), "site_id": r[0], "name": r[1], "district": r[2],
         "developer": r[3], "obj_class": r[4], "weighted": round(r[7], 1), "rank": r[8]}
        for d, r in nearby[:10]
    ]

    out = {
        "generated_at": __import__("datetime").datetime.now().isoformat(timespec="seconds"),
        "parcel": parcel,
        "n_compared_jk": stats["n_jk"],
        "weighted_score_distribution": stats,
        "component_comparison": comp_stats,
        "weights_used": {"education":0.30,"health":0.15,"retail":0.20,"transit":0.20,"leisure":0.15},
        "top10_best_jk_ekb": top10,
        "10_closest_jk_to_parcel": closest,
    }

    json_path = OUT / "parcel_66_41_0204016_10.json"
    with open(json_path, "w") as f:
        json.dump(out, f, ensure_ascii=False, indent=2, default=str)
    print(f"JSON: {json_path}")

    # Build HTML
    html = build_html(out)
    html_path = OUT / "parcel_66_41_0204016_10.html"
    with open(html_path, "w") as f:
        f.write(html)
    print(f"HTML: {html_path}")

    conn.close()

def econ_block(e):
    if not e:
        return "<p class=muted>Нет данных Объектива для этого района.</p>"
    f = lambda x, suf="": (f"{x:.1f}{suf}" if isinstance(x,(int,float)) else "—")
    method_note = "по совпадению имени ЖК" if e["district_method"]=="name_match" \
                  else f"по голосованию 5 ближайших ЖК (расст. до репрезентанта {e['district_dist_m']:.0f} м)"
    return f"""
<table>
<tr><th>Район (Объектив)</th><td><b>{e["district"]}</b> <span class=muted>({method_note})</span></td></tr>
<tr><th>Проектов в районе</th><td>{e["n_projects"]}</td></tr>
<tr><th>Лотов всего / продано</th><td>{e["n_lots"] or 0:,} / <b>{e["n_sold"] or 0:,}</b> ({f(e["sold_pct"], '%')})</td></tr>
<tr><th>Цена за м² (медиана)</th><td><b>{f(e["median_price_m2"])} тыс ₽</b> · P25={f(e["p25_price_m2"])} · P75={f(e["p75_price_m2"])}</td></tr>
<tr><th>Средняя площадь сделки</th><td>{f(e["avg_area_sold_m2"])} м²</td></tr>
<tr><th>Скорость продаж (real)</th><td><b>{f(e["velocity_per_month"])}</b> зарег. ДДУ/корпус/мес <span class=muted>(по 12 мес)</span></td></tr>
<tr><th>Средняя готовность</th><td>{f(e["avg_readiness_pct"], '%')}</td></tr>
<tr><th>Цена corp_sum (взвеш.)</th><td class=muted>{f(e["weighted_price_m2_corp_sum"])} тыс ₽/м² · скорость {f(e["deals_per_month_corp_sum"])}</td></tr>
<tr><th>Распродажа стока (corp_sum)</th><td class=muted>{f(e["months_to_sellout"], ' мес')}</td></tr>
</table>
<p class=muted><b>real_*</b> рассчитаны по {e["n_lots"] or 0:,} лотам из Поквартирные/Лоты (303 677 квартир Екб). Это per-flat, основной источник правды.</p>
"""

def build_html(d):
    p = d["parcel"]
    cs = d["component_comparison"]
    poi_table_rows = []
    for cat, items in p["nearest_pois"].items():
        nearest = items[0] if items else None
        poi_table_rows.append(
            f"<tr><td>{cat}</td>"
            f"<td>{nearest['name'] if nearest else '—'}</td>"
            f"<td class='r'>{('%.0f м' % nearest['distance_m']) if nearest else '—'}</td>"
            f"<td class='r'>{p['features'].get(f'{cat}_count_500m','')}</td>"
            f"<td class='r'>{p['features'].get(f'{cat}_count_1km','')}</td></tr>"
        )

    comp_rows = []
    for c, v in cs.items():
        delta = v["parcel"] - v["median_jk"]
        cls = "g" if delta > 0 else ("r" if delta < 0 else "")
        comp_rows.append(
            f"<tr><td>{c}</td><td class='r'>{v['parcel']}</td>"
            f"<td class='r'>{v['median_jk']}</td><td class='r'>{v['p75_jk']}</td>"
            f"<td class='r {cls}'>{'+' if delta>=0 else ''}{delta:.1f}</td></tr>"
        )

    closest_rows = "".join(
        f"<tr><td class='r'>{x['distance_m']:.0f} м</td>"
        f"<td>{x['name'] or '—'}</td><td>{x['district'] or '—'}</td>"
        f"<td>{x['developer'] or ''}</td><td>{x['obj_class'] or ''}</td>"
        f"<td class='r'>{x['weighted']}</td><td class='r'>#{x['rank']}</td></tr>"
        for x in d["10_closest_jk_to_parcel"]
    )

    top_rows = "".join(
        f"<tr><td class='r'>#{x['rank']}</td><td>{x['name'] or '—'}</td>"
        f"<td>{x['district'] or '—'}</td><td>{x['developer'] or ''}</td>"
        f"<td>{x['obj_class'] or ''}</td><td class='r'>{x['weighted']:.1f}</td></tr>"
        for x in d["top10_best_jk_ekb"]
    )

    return f"""<!doctype html>
<html lang=ru><head><meta charset=utf-8>
<title>Анализ участка {p['name']}</title>
<style>
body{{font-family:-apple-system,Segoe UI,sans-serif;max-width:980px;margin:32px auto;padding:0 16px;color:#222;line-height:1.5}}
h1{{font-size:24px;margin:0 0 8px}}
h2{{font-size:18px;margin-top:32px;border-bottom:1px solid #eee;padding-bottom:6px}}
.kpi{{display:grid;grid-template-columns:repeat(5,1fr);gap:12px;margin:18px 0}}
.kpi div{{background:#f5f7fa;border-radius:8px;padding:14px}}
.kpi b{{display:block;font-size:11px;color:#666;text-transform:uppercase;margin-bottom:4px}}
.kpi span{{font-size:24px;font-weight:600}}
.big{{font-size:42px;font-weight:700;color:#0a6}}
table{{border-collapse:collapse;width:100%;margin:8px 0;font-size:13px}}
th,td{{border:1px solid #ddd;padding:6px 10px;text-align:left}}
th{{background:#f5f7fa;font-weight:600}}
.r{{text-align:right}}
.g{{color:#0a6;font-weight:600}}
.r.r{{color:#c33;font-weight:600}}
.muted{{color:#888;font-size:12px}}
.note{{background:#fffbe6;border-left:4px solid #f0c000;padding:10px 14px;border-radius:4px;margin:12px 0}}
</style></head><body>
<h1>Анализ участка <code>66:41:0204016:10</code></h1>
<div class=muted>Сгенерировано {d["generated_at"]} · Сравнение с {d["n_compared_jk"]} строящимися ЖК Екатеринбурга</div>

<div class="note">
  Координаты участка получены из ссылки на NSPD-карту (EPSG:3857 → WGS84):
  <b>{p['lat']}, {p['lon']}</b>. Район — центрально-северная часть ЕКБ (Пионерский / Втузгородок).
</div>

<div class=kpi>
<div><b>Итоговый балл</b><span class=big>{p["weighted_total"]:.1f}</span><span class=muted>из 100</span></div>
<div><b>Ранг по ЕКБ</b><span>#{p["rank_overall"]} / {d["n_compared_jk"]+1}</span></div>
<div><b>Перцентиль</b><span>{(1 - (p["rank_overall"]-1)/(d["n_compared_jk"]+1))*100:.0f}%</span></div>
<div><b>Медиана ЖК</b><span>{d["weighted_score_distribution"]["median"]}</span></div>
<div><b>Топ-25% ЖК ≥</b><span>{d["weighted_score_distribution"]["p75"]}</span></div>
</div>

<h2>Компоненты (взвешенные)</h2>
<table><thead><tr><th>Компонент</th><th class=r>Участок</th><th class=r>Медиана ЖК</th><th class=r>P75 ЖК</th><th class=r>Δ vs медиана</th></tr></thead>
<tbody>{"".join(comp_rows)}</tbody></table>
<div class=muted>Веса: образование 20% · здоровье 10% · ритейл 15% · транспорт 15% · досуг 10% · экономика 30%</div>

<h2>Экономика района (Объектив API, последние 90 дней)</h2>
{econ_block(p.get("economics"))}

<h2>Ближайшие POI вокруг участка</h2>
<table><thead><tr><th>Категория</th><th>Ближайший</th><th class=r>До него</th><th class=r>В 500 м</th><th class=r>В 1 км</th></tr></thead>
<tbody>{"".join(poi_table_rows)}</tbody></table>

<h2>10 ближайших ЖК (для прямого бенчмарка)</h2>
<table><thead><tr><th class=r>Расст.</th><th>Название</th><th>Район</th><th>Девелопер</th><th>Класс</th><th class=r>Балл</th><th class=r>Ранг</th></tr></thead>
<tbody>{closest_rows}</tbody></table>

<h2>Топ-10 ЖК ЕКБ по локационной привлекательности</h2>
<table><thead><tr><th class=r>Ранг</th><th>Название</th><th>Район</th><th>Девелопер</th><th>Класс</th><th class=r>Балл</th></tr></thead>
<tbody>{top_rows}</tbody></table>

<h2>Методика</h2>
<p><b>Источник POI:</b> OpenStreetMap (Overpass API), bbox по всем 381 ЖК Свердл.</p>
<p><b>Логика:</b> для каждой категории — расстояние-в-балл (piecewise linear от <i>ideal_m</i> к <i>max_m</i>),
далее агрегация в 5 компонент с весами (max-pool там, где категории альтернативны: ритейл = max(big, 0.7×med); транспорт = max(metro, 0.85×tram, 0.7×bus)).
Финальный балл — взвешенная сумма компонент.</p>
<p><b>База данных:</b> локальная SQLite <code>analysis.db</code> (sites/pois/features/scores), под-выборка строящихся ЖК ЕКБ из прода <code>domrf_kn_objects</code>.</p>
</body></html>"""

if __name__ == "__main__":
    main()