#!/usr/bin/env python3
"""
各渠道付费用户近14天完课频次差异分析。
限定：2026年3月后订单、剔除退费。
"""
import psycopg2
import psycopg2.extras
from collections import defaultdict
import numpy as np

conn = psycopg2.connect(
    host="bj-postgres-16pob4sg.sql.tencentcdb.com",
    port=28591,
    user="ai_member",
    password="LdfjdjL83h3h3^$&**YGG*",
    dbname="vala_bi"
)
conn.set_session(autocommit=True)
cur = conn.cursor(cursor_factory=psycopg2.extras.DictCursor)

START = "2026-05-09"
END   = "2026-05-22"

print("=" * 80)
print("📊 各渠道付费用户近14天完课频次差异")
print("   限定：3月后订单 + 剔除退费")
print("=" * 80)

# ── 渠道分类 ─────────────────────────────────────────────────
ENDPOINT_INTERNAL = {'app-active-h5-0-0', 'app-sales-bj-qhm-0'}
def classify(kf):
    if kf in ENDPOINT_INTERNAL: return "端内"
    if kf.startswith("sales-adp"): return "销售渠道"
    if kf == "newmedia-dianpu-xhs-0-0": return "小红书店铺"
    if kf.startswith("newmedia-daren"): return "达人直播"
    if kf == "newmedia-dianpu-wwxx-0-0": return "万物"
    return "其他端外"

# ═══════════════════════════════════════════════════════════
# Step 1: 付费用户 + 渠道 + 等级
# ═══════════════════════════════════════════════════════════
print("\n[1/4] 获取付费用户...")
cur.execute("""
    SELECT o.account_id, o.goods_id, o.key_from
    FROM bi_vala_order o
    INNER JOIN bi_vala_app_account a ON o.account_id = a.id AND a.status = 1
    WHERE o.order_status = 3 AND o.pay_success_date >= '2026-03-01' AND o.deleted_at IS NULL
""")
rows = cur.fetchall()

user_goods = defaultdict(set)
user_channels = defaultdict(set)
for r in rows:
    user_goods[r[0]].add(r[1])
    user_channels[r[0]].add(classify(r[2]))

L1_G = {57,60,63}; L2_G = {31,32,33,54}; L12_G = {61}
def level(aid):
    g = user_goods.get(aid, set())
    h1, h2, h12 = bool(g & L1_G), bool(g & L2_G), bool(g & L12_G)
    if h12 or (h1 and h2): return "L1+L2"
    if h1: return "仅L1"
    if h2: return "仅L2"
    return "其他"

# ═══════════════════════════════════════════════════════════
# Step 2: 角色映射
# ═══════════════════════════════════════════════════════════
print("[2/4] 获取角色映射...")
all_aids = list(user_goods.keys())
cur.execute("SELECT id, account_id FROM bi_vala_app_character WHERE account_id = ANY(%s) AND deleted_at IS NULL", (all_aids,))
char_to_account = {}
for r in cur.fetchall():
    char_to_account[r[0]] = r[1]
all_char_ids = set(char_to_account.keys())

# ═══════════════════════════════════════════════════════════
# Step 3: 各角色近14天完课次数
# ═══════════════════════════════════════════════════════════
print("[3/4] 统计各角色近14天完课次数（8张分表）...")
char_completion_count = defaultdict(int)
for shard in range(8):
    cur.execute(f"""
        SELECT user_id, COUNT(*) as cnt
        FROM bi_user_chapter_play_record_{shard}
        WHERE play_status = 1
          AND updated_at >= %s AND updated_at < %s::date + interval '1 day'
        GROUP BY user_id
    """, (START, END))
    for r in cur.fetchall():
        cid = r[0]
        if cid in all_char_ids:
            char_completion_count[cid] += r[1]

# 汇总到 account
account_completions = defaultdict(int)
for cid, cnt in char_completion_count.items():
    account_completions[char_to_account[cid]] += cnt

# ═══════════════════════════════════════════════════════════
# Step 4: 按渠道聚合统计
# ═══════════════════════════════════════════════════════════
print("[4/4] 按渠道聚合...")

CHANNELS = ["端内", "销售渠道", "小红书店铺", "达人直播", "万物", "其他端外"]
channel_user_set = defaultdict(set)
for aid, chs in user_channels.items():
    for ch in chs:
        channel_user_set[ch].add(aid)

def stats(aid_set):
    """返回 (总人数, 有完课人数, 总完课次数, 完课次数列表)"""
    comps = []
    active = 0
    for aid in aid_set:
        c = account_completions.get(aid, 0)
        if c > 0:
            active += 1
            comps.append(c)
    total_comps = sum(comps)
    return len(aid_set), active, total_comps, comps

# ── 表格输出 ──
header = f"{'渠道':<12s} {'等级':<8s} {'总付费':>6s} {'活跃人数':>7s} {'活跃率':>7s} {'总完课次':>7s} {'人均(全)':>8s} {'人均(活跃)':>9s} {'中位数':>6s} {'P75':>5s} {'P90':>5s}"
print(f"\n{header}")
print("-" * 105)

for ch in CHANNELS:
    users = channel_user_set.get(ch, set())
    if not users:
        continue
    # 渠道总计
    tot, act, total_c, comps = stats(users)
    avg_all = total_c / tot if tot else 0
    avg_act = total_c / act if act else 0
    sorted_c = sorted(comps)
    med = np.median(sorted_c) if sorted_c else 0
    p75 = np.percentile(sorted_c, 75) if sorted_c else 0
    p90 = np.percentile(sorted_c, 90) if sorted_c else 0
    print(f"{ch:<12s} {'合计':<8s} {tot:>6d} {act:>7d} {act/tot*100:>6.1f}% {total_c:>7d} {avg_all:>8.1f} {avg_act:>9.1f} {med:>6.0f} {p75:>5.0f} {p90:>5.0f}")

    # 按等级拆分
    for lv in ["仅L1", "仅L2", "L1+L2"]:
        subset = {a for a in users if level(a) == lv}
        if not subset:
            continue
        t, a, tc, cs = stats(subset)
        aa = tc / t if t else 0
        ag = tc / a if a else 0
        sc = sorted(cs)
        md = np.median(sc) if sc else 0
        p7 = np.percentile(sc, 75) if sc else 0
        p9 = np.percentile(sc, 90) if sc else 0
        print(f"{'':<12s} {lv:<8s} {t:>6d} {a:>7d} {a/t*100:>6.1f}% {tc:>7d} {aa:>8.1f} {ag:>9.1f} {md:>6.0f} {p7:>5.0f} {p9:>5.0f}")
    print("-" * 105)

# ── 频次分布 ──
print(f"\n{'═' * 70}")
print("📋 完课频次分布（活跃用户）")
print(f"{'═' * 70}")
BINS = [(1,1), (2,2), (3,4), (5,7), (8,14), (15, 30), (31, 999)]

def dist(aid_set):
    comps = [account_completions.get(a, 0) for a in aid_set if account_completions.get(a, 0) > 0]
    total = len(comps)
    d = []
    for lo, hi in BINS:
        cnt = sum(1 for c in comps if lo <= c <= hi)
        pct = cnt / total * 100 if total else 0
        d.append((f"{lo}-{hi}" if hi < 999 else f"{lo}+", cnt, pct))
    return d, total

print(f"\n{'渠道':<12s} {'活跃':>5s}", end="")
for lo, hi in BINS:
    label = f"{lo}-{hi}" if hi < 999 else f"{lo}+"
    print(f" {label:>7s}", end="")
print()

for ch in CHANNELS:
    users = channel_user_set.get(ch, set())
    if not users:
        continue
    d, tot = dist(users)
    if tot == 0:
        continue
    print(f"{ch:<12s} {tot:>5d}", end="")
    for label, cnt, pct in d:
        print(f" {pct:>6.1f}%", end="")
    print()

# ── 频道对比柱状图 ──
print(f"\n{'═' * 70}")
print("📊 人均完课次数对比（活跃用户）")
print(f"{'═' * 70}")
for ch in CHANNELS:
    users = channel_user_set.get(ch, set())
    if not users:
        continue
    _, act, tc, _ = stats(users)
    avg = tc / act if act else 0
    bar = "█" * max(1, int(avg * 2))
    print(f"  {ch:<12s} 人均 {avg:5.1f} 次  {bar}")

cur.close()
conn.close()
print("\n✅ 完成")