微信小程序页面迁移校验之前 P5任务处理之前

scripts/ops/_diagnose_member_visit_conflict.py

@@ -0,0 +1,84 @@
+"""诊断 DWS_MEMBER_VISIT 唯一约束冲突的具体 order"""
+import os
+from pathlib import Path
+from dotenv import load_dotenv
+
+load_dotenv(Path(__file__).resolve().parents[2] / ".env")
+
+PG_DSN = os.environ.get("PG_DSN")
+if not PG_DSN:
+    raise RuntimeError("PG_DSN 未设置")
+
+import psycopg2
+import psycopg2.extras
+
+conn = psycopg2.connect(PG_DSN)
+conn.autocommit = True
+
+def q(sql):
+    with conn.cursor(cursor_factory=psycopg2.extras.RealDictCursor) as cur:
+        cur.execute(sql)
+        return cur.fetchall()
+
+print("DWS_MEMBER_VISIT 冲突 order 详情")
+print("=" * 60)
+
+# 找出旧数据中与新窗口重叠的 order 的详细信息
+rows = q("""
+    WITH old_visits AS (
+        SELECT site_id, member_id, order_settle_id, visit_date
+        FROM dws.dws_member_visit_detail
+        WHERE visit_date < '2025-11-01'
+    ),
+    new_source AS (
+        SELECT DISTINCT site_id, member_id, order_settle_id, 
+               pay_time,
+               (CASE WHEN EXTRACT(HOUR FROM pay_time) < 8 
+                     THEN (pay_time - INTERVAL '1 day')::date 
+                     ELSE pay_time::date END) AS biz_date
+        FROM dwd.dwd_settlement_head
+        WHERE member_id IS NOT NULL AND member_id != 0
+          AND pay_time >= '2025-11-01' AND pay_time <= '2026-02-28'
+    )
+    SELECT 
+        o.site_id, o.member_id, o.order_settle_id,
+        o.visit_date AS old_visit_date,
+        n.biz_date AS new_biz_date,
+        n.pay_time
+    FROM old_visits o
+    JOIN new_source n ON o.site_id = n.site_id 
+        AND o.member_id = n.member_id 
+        AND o.order_settle_id = n.order_settle_id
+    ORDER BY o.visit_date
+""")
+
+print(f"冲突 order 数: {len(rows)}")
+for r in rows:
+    print(f"  order={r['order_settle_id']}, member={r['member_id']}")
+    print(f"    旧 visit_date={r['old_visit_date']}, 新 biz_date={r['new_biz_date']}, pay_time={r['pay_time']}")
+
+# 检查这些 order 的 pay_time 是否在 biz_date 边界上
+print()
+print("分析: 这些 order 的 pay_time 在凌晨（< 8:00），")
+print("biz_date 归属到前一天，但 DWD 中的数据可能在不同时间被处理")
+
+# 检查 dwd_settlement_head 中这些 order 的 SCD2 版本
+if rows:
+    order_ids = [r['order_settle_id'] for r in rows]
+    placeholders = ','.join(['%s'] * len(order_ids))
+    rows2 = q(f"""
+        SELECT order_settle_id, member_id, pay_time, 
+               created_at, updated_at,
+               scd2_valid_from, scd2_valid_to, scd2_is_current
+        FROM dwd.dwd_settlement_head
+        WHERE order_settle_id IN ({placeholders})
+    """.replace('%s', ','.join(str(x) for x in order_ids)))
+    
+    print(f"\n这些 order 在 dwd_settlement_head 中的记录:")
+    for r in rows2:
+        print(f"  order={r['order_settle_id']}, member={r['member_id']}")
+        print(f"    pay_time={r['pay_time']}, scd2_current={r['scd2_is_current']}")
+        print(f"    scd2_from={r['scd2_valid_from']}, scd2_to={r['scd2_valid_to']}")
+
+conn.close()
+print("\n诊断完成。")