"""诊断：NULL level_name 助教的 SCD2 最早记录 vs daily 最早日期"""
import os, sys
from pathlib import Path
from dotenv import load_dotenv

load_dotenv(Path(__file__).resolve().parents[2] / ".env")

import psycopg2
import psycopg2.extras

dsn = os.environ.get("PG_DSN")
if not dsn:
    raise RuntimeError("PG_DSN 未设置")

conn = psycopg2.connect(dsn)
cur = conn.cursor(cursor_factory=psycopg2.extras.RealDictCursor)

# 对比每个 NULL assistant_id 的 SCD2 最早时间 vs daily 最早日期
cur.execute("""
    WITH null_aids AS (
        SELECT DISTINCT assistant_id
        FROM dws.dws_assistant_daily_detail
        WHERE assistant_level_name IS NULL
    ),
    scd2_earliest AS (
        SELECT da.assistant_id, MIN(da.scd2_start_time) AS earliest_scd2
        FROM dwd.dim_assistant da
        JOIN null_aids n ON da.assistant_id = n.assistant_id
        GROUP BY da.assistant_id
    ),
    daily_earliest AS (
        SELECT d.assistant_id, MIN(d.stat_date) AS earliest_daily
        FROM dws.dws_assistant_daily_detail d
        JOIN null_aids n ON d.assistant_id = n.assistant_id
        WHERE d.assistant_level_name IS NULL
        GROUP BY d.assistant_id
    )
    SELECT 
        de.assistant_id,
        de.earliest_daily,
        se.earliest_scd2,
        se.earliest_scd2::date AS scd2_date,
        CASE WHEN de.earliest_daily < se.earliest_scd2::date THEN 'DAILY_BEFORE_SCD2'
             ELSE 'SCD2_COVERS' END AS status
    FROM daily_earliest de
    LEFT JOIN scd2_earliest se ON de.assistant_id = se.assistant_id
    ORDER BY de.earliest_daily
""")
rows = cur.fetchall()
print(f"=== {len(rows)} 个 NULL level_name 助教 ===")
before_count = 0
for r in rows:
    status = r['status']
    if status == 'DAILY_BEFORE_SCD2':
        before_count += 1
    print(f"  aid={r['assistant_id']}, daily_earliest={r['earliest_daily']}, "
          f"scd2_earliest={r['scd2_date']}, status={status}")

print(f"\n总计: {before_count}/{len(rows)} 个助教的 daily 数据早于 SCD2 首条记录")

cur.close()
conn.close()