#!/usr/bin/env python3
"""查询2026年3月后未退费订单，激活课程在下单30天内的学习进度分布"""
import psycopg2
import os
import sys

# 数据库连接
conn = psycopg2.connect(
    host="bj-postgres-16pob4sg.sql.tencentcdb.com",
    port=28591,
    user="ai_member",
    password="LdfjdjL83h3h3^$&**YGG*",
    dbname="vala_bi"
)

cur = conn.cursor()

sql = """WITH orders AS (
    SELECT o.id, o.out_trade_no, o.account_id, o.pay_success_date
    FROM bi_vala_order o
    JOIN bi_vala_app_account a ON o.account_id = a.id
    WHERE o.pay_success_date >= '2026-03-01'
      AND o.order_status != 4
      AND o.pay_success_date IS NOT NULL
      AND a.status = 1
),
tickets AS (
    SELECT o.out_trade_no, o.account_id, o.pay_success_date,
           t.character_id, t.season_package_level
    FROM orders o
    JOIN bi_vala_seasonal_ticket t ON o.out_trade_no = t.out_trade_no
    WHERE t.status = 1 
      AND t.deleted_at IS NULL
      AND t.season_package_level IN ('A1', 'A2')
),
all_chapter_records AS (
    SELECT user_id, chapter_id, created_at FROM bi_user_chapter_play_record_0 WHERE play_status = 1
    UNION ALL
    SELECT user_id, chapter_id, created_at FROM bi_user_chapter_play_record_1 WHERE play_status = 1
    UNION ALL
    SELECT user_id, chapter_id, created_at FROM bi_user_chapter_play_record_2 WHERE play_status = 1
    UNION ALL
    SELECT user_id, chapter_id, created_at FROM bi_user_chapter_play_record_3 WHERE play_status = 1
    UNION ALL
    SELECT user_id, chapter_id, created_at FROM bi_user_chapter_play_record_4 WHERE play_status = 1
    UNION ALL
    SELECT user_id, chapter_id, created_at FROM bi_user_chapter_play_record_5 WHERE play_status = 1
    UNION ALL
    SELECT user_id, chapter_id, created_at FROM bi_user_chapter_play_record_6 WHERE play_status = 1
    UNION ALL
    SELECT user_id, chapter_id, created_at FROM bi_user_chapter_play_record_7 WHERE play_status = 1
),
level_records AS (
    SELECT pr.user_id, pr.chapter_id, pr.created_at, l.course_level
    FROM all_chapter_records pr
    JOIN bi_level_unit_lesson l ON pr.chapter_id = l.id
    WHERE l.course_level IN ('L1', 'L2')
),
ticket_progress AS (
    SELECT 
        t.out_trade_no,
        t.account_id,
        t.character_id,
        t.season_package_level,
        t.pay_success_date,
        COUNT(DISTINCT lr.chapter_id) AS completed_lessons
    FROM tickets t
    LEFT JOIN level_records lr 
        ON t.character_id = lr.user_id
        AND lr.course_level = CASE 
            WHEN t.season_package_level = 'A1' THEN 'L1' 
            WHEN t.season_package_level = 'A2' THEN 'L2' 
        END
        AND lr.created_at >= t.pay_success_date
        AND lr.created_at <= t.pay_success_date + INTERVAL '30 days'
    GROUP BY t.out_trade_no, t.account_id, t.character_id, t.season_package_level, t.pay_success_date
)
SELECT 
    CASE WHEN season_package_level = 'A1' THEN 'L1' ELSE 'L2' END AS course_level,
    completed_lessons,
    COUNT(*) AS ticket_count,
    ROUND(COUNT(*) * 100.0 / SUM(COUNT(*)) OVER (PARTITION BY season_package_level), 1) AS pct
FROM ticket_progress
GROUP BY season_package_level, completed_lessons
ORDER BY season_package_level, completed_lessons;"""

print("正在查询...")
cur.execute(sql)
rows = cur.fetchall()

print(f"\n{'='*60}")
print("2026年3月后未退费订单 → 激活课程下单30天内学习进度分布")
print(f"{'='*60}")

l1_rows = [r for r in rows if r[0] == 'L1']
l2_rows = [r for r in rows if r[0] == 'L2']

for label, data in [('L1', l1_rows), ('L2', l2_rows)]:
    print(f"\n--- {label} ---")
    total = sum(r[2] for r in data)
    print(f"总 ticket 数: {total}")
    print(f"{'完课数':>8} | {'ticket数':>10} | {'占比':>8}")
    print("-" * 35)
    for r in data:
        print(f"{r[1]:>8} | {r[2]:>10} | {r[3]:>7}%")
    
    # 统计摘要
    if data:
        completed_values = []
        for r in data:
            completed_values.extend([r[1]] * r[2])
        import statistics
        avg = sum(completed_values) / len(completed_values) if completed_values else 0
        median = statistics.median(completed_values) if completed_values else 0
        zero_pct = sum(1 for v in completed_values if v == 0) / len(completed_values) * 100 if completed_values else 0
        print(f"\n摘要: 平均={avg:.1f}节, 中位数={median:.0f}节, 0节课占比={zero_pct:.1f}%")

cur.close()
conn.close()
print("\n查询完成。")