🤖 每日自动备份 - 2026-05-07 08:00:01

2026-05-07 08:00:01 +08:00 · 2026-05-07 08:00:01 +08:00 · 968286d9c9
commit 968286d9c9
parent 6510dc41ca
17 changed files with 1526 additions and 112 deletions
--- a/.vala_skill_hashes
+++ b/.vala_skill_hashes
@ -12,4 +12,4 @@ vala-component-practice-stat 8e768e2641019d27bd41f4647d2d90f24182a0554dad5ad9f41
 cron-schedule e103cbb1806b28c891b9c856963325086ecaff32edec208f0a841865f26e8f3e
 refund-user-learning-analysis 648fd4ae2b29167fd66eab4245bdaaef00242db3131f4919cc02f07ca2a9b59c
 phone-chapter-query ac429b4da5a89db16efdf1066edf4ecb1c050b93aff20dd4c652af5f5568e44f
-vala-order-amortization-stat de5f77b9e4ab865df5efe595e59e877257ac489a195fb969b9d20d1a71158b97
+vala-order-amortization-stat fd9da99f313347aa11a4cad8e5d0c9226eac8649b45db94dfa60e37015ab7491
--- a/memory/.dreams/short-term-recall.json
+++ b/memory/.dreams/short-term-recall.json
@ -0,0 +1,37 @@
+{
+  "version": 1,
+  "updatedAt": "2026-05-06T12:28:28.905Z",
+  "entries": {
+    "memory:memory/2026-05-06.md:1:20": {
+      "key": "memory:memory/2026-05-06.md:1:20",
+      "path": "memory/2026-05-06.md",
+      "startLine": 1,
+      "endLine": 20,
+      "source": "memory",
+      "snippet": "# 2026-05-06 工作日志 ## 任务1：学而思/讯飞/希沃/京东方月度渠道订单报表 - [李承龙需求] 创建按月汇总四个渠道订单的 Python 脚本 - 脚本路径：`/root/.openclaw/workspace/scripts/monthly_channel_order_report.py` - **渠道映射：** 仅端内 `key_from='app-active-h5-0-0'`，sale_channel: 21=学而思, 22=讯飞, 26=希沃, 27=京东方 - **输出格式：** 四个渠道各一个 Sheet，每月按天展示（学而思/希沃/京东方只显示有订单日期，讯飞显示全月每天） - **指标差异：** - 学而思/希沃/京东方：订单数、GMV、CPS分成（学而思×0.45，希沃×0.4，京东方×0.3）、订单号（分号拼接） - 讯飞：订单数、GMV、新增注册（download_channel LIKE '%讯飞%'）、CPA分成（新增注册×7）、订单号 - **退费剔除：** NOT EXISTS (bi_refund_order.status=3) AND order_status != 4 - **测试运行：** 2026-04 月份，新增注册 537（讯飞），订单合计 15 单，GMV 14585 元 - **订单号格式问题：** Excel 科学计数法 → 设置单元格为文本格式解决 - **psycopg2 转义：** LIKE 中 `%` 需写成 `%%`（psycopg2 的 `%s` 占位符冲突） ## 任务2：销售线",
+      "recallCount": 1,
+      "dailyCount": 0,
+      "groundedCount": 0,
+      "totalScore": 1,
+      "maxScore": 1,
+      "firstRecalledAt": "2026-05-06T12:28:28.905Z",
+      "lastRecalledAt": "2026-05-06T12:28:28.905Z",
+      "queryHashes": [
+        "c30fa9d13974"
+      ],
+      "recallDays": [
+        "2026-05-06"
+      ],
+      "conceptTags": [
+        "学而思/讯飞/希沃/京东方月度渠道订单报表",
+        "key-from",
+        "app-active-h5-0-0",
+        "sale-channel",
+        "学而思/希沃/京东方只显示有订单日期",
+        "学而思/希沃/京东方",
+        "0.45",
+        "0.4"
+      ]
+    }
+  }
+}
--- a/memory/2026-05-06-1025.md
+++ b/memory/2026-05-06-1025.md
@ -0,0 +1,5 @@
+# Session: 2026-05-06 10:25:14 GMT+8
+
+- **Session Key**: agent:main:feishu:direct:ou_e63ce6b760ad39382852472f28fbe2a2
+- **Session ID**: 58b30963-b037-4f50-8b15-04e77ad8fef0
+- **Source**: feishu
--- a/memory/2026-05-06-order-amortization-sql.md
+++ b/memory/2026-05-06-order-amortization-sql.md
@ -0,0 +1,53 @@
+# Session: 2026-05-06 16:53:32 GMT+8
+
+- **Session Key**: agent:main:feishu:direct:ou_e63ce6b760ad39382852472f28fbe2a2
+- **Session ID**: 690e7978-4b15-4ce8-bfa3-9f8bb9271428
+- **Source**: feishu
+
+## Conversation Summary
+
+assistant: 让我尝试用其他方式读取这张图片：
+assistant: OCR 识别不太完整。让我尝试提取为文本再做更好的识别：
+assistant: OCR 效果有限，但结合上下文我能推断这个表格的结构。让我换个思路，下载图片后用 Python OCR：
+assistant: 承龙，OCR 识别效果不理想，表格的汉字内容没能完全准确提取。让我基于已识别到的内容和你之前的需求上下文，整理一下我理解的表格结构，请你确认是否正确：
+
+根据 OCR 结果和上下文，这张表格应该是一个 **3行×3列** 的结构：
+
+| | 新增（金额） | 退费（金额） |
+|---|---|---|
+| **上月订单转正** | 下单时间在上个月且在上个月未转正的订单金额总和，剔除上个月全部退费的订单，部分退费只计算有效金额，不考虑本月退费情况 | 这些订单在本月的退费金额 |
+| **本月订单转正** | 下单时间在本月且在本月已经转正的订单的金额总和，不考虑退费情况 | 这些订单在本月的退费金额 |
+| **本月订单未转正** | 下单时间在本月且在本月未转正的订单的金额总和，不考虑退费情况 | 这些订单在本月的退费金额 |
+
+也就是：
+- 行：上月订单转正、本月订单转正、本月订单未转正
+- 列：新增（即预收账款金额）、退费（这些订单在本账期的退费金额）
+
+如果理解正确，我需要确认几个问题：
+
+1. **"新增"列中的金额**，是订单的**含税金额（pay_amount）**还是**税后金额**？还是**退后金额（pay_amount - 退费）**？
+2. **"退费"列**是指这些订单在本账期内发生的退费金额对吧？
+3. 表格里的"上月"是相对于账期的（如 3 月账期，"上月"就是 2 月），"本月"就是账期所在月（3 月），对吗？
+user: 1、是指订单的含税金额
+2、对的
+3、对的
+assistant: 好的，明确了。Sheet3 "本月预收账款" 结构如下：
+
+| | 新增（含税金额） | 退费 |
+|---|---|---|
+| **上月订单转正** | 上月下单+上月未转正+本月转正的订单；剔除上月已全额退费的；部分退费只算有效金额(pay - 退费)；不考虑本月退费 | 这些订单在本月的退费金额 |
+| **本月订单转正** | 本月下单+本月已转正的订单金额总和；不考虑退费 | 这些订单在本月的退费金额 |
+| **本月订单未转正** | 本月下单+本月未转正的订单金额总和；不考虑退费 | 这些订单在本月的退费金额 |
+
+现在写 SQL。相对账期 "本月"=账期月，"上月"=账期月前一个月。需要计算这三个维度的新增和退费。
+assistant: 现在修改 `run.py`，增加 Sheet3 的支持：
+assistant: 现在跑 3 月数据测试：
+assistant: 运行成功。验证一下 Sheet3 的内容：
+assistant: Sheet3 数据跑出来了。发送文件：
+assistant: 文件已发送。三个 Sheet 分别是：汇总表、订单明细、本月预收账款。看看数据有没有问题？
+user: [OpenClaw heartbeat poll]
+assistant: `vala-order-amortization-stat` 有变更已推送成功。需要通知李若松。
+assistant: 好的，`vala-order-amortization-stat` 已推送到 SkillHub，需要通知李若松。李若松的 user_id 是 `4aagb443`。
+assistant: 已通知李若松。
+
+HEARTBEAT_OK
--- a/memory/2026-05-06.md
+++ b/memory/2026-05-06.md
@ -0,0 +1,36 @@
+
+# 2026-05-06 工作日志
+
+## 任务1：学而思/讯飞/希沃/京东方月度渠道订单报表
+
+- [李承龙需求] 创建按月汇总四个渠道订单的 Python 脚本
+- 脚本路径：`/root/.openclaw/workspace/scripts/monthly_channel_order_report.py`
+- **渠道映射：** 仅端内 `key_from='app-active-h5-0-0'`，sale_channel: 21=学而思, 22=讯飞, 26=希沃, 27=京东方
+- **输出格式：** 四个渠道各一个 Sheet，每月按天展示（学而思/希沃/京东方只显示有订单日期，讯飞显示全月每天）
+- **指标差异：**
+  - 学而思/希沃/京东方：订单数、GMV、CPS分成（学而思×0.45，希沃×0.4，京东方×0.3）、订单号（分号拼接）
+  - 讯飞：订单数、GMV、新增注册（download_channel LIKE '%讯飞%'）、CPA分成（新增注册×7）、订单号
+- **退费剔除：** NOT EXISTS (bi_refund_order.status=3) AND order_status != 4
+- **测试运行：** 2026-04 月份，新增注册 537（讯飞），订单合计 15 单，GMV 14585 元
+- **订单号格式问题：** Excel 科学计数法 → 设置单元格为文本格式解决
+- **psycopg2 转义：** LIKE 中 `%` 需写成 `%%`（psycopg2 的 `%s` 占位符冲突）
+
+## 任务2：销售线索用户分析报表
+
+- [李承龙需求] 输入包含462个用户ID的Excel，输出每个用户+角色的完整画像报表
+- 脚本路径：`/root/.openclaw/workspace/scripts/lead_user_analysis.py`
+- **重要口径修正（李承龙确认）：**
+  - **体验课定义：** 固定为 L1 U00(L01-L05) + L2 U00(L01-L05) 共10节课，**不通过 expire_time 区分**
+  - L1 chapter_id: 343, 344, 345, 346, 348
+  - L2 chapter_id: 55, 56, 57, 58, 59
+  - **完成时间：** 取 `bi_user_chapter_play_record` 中 `play_status=1` 记录的**最早 `updated_at`**（非 `created_at`）
+- **报表列：** 用户ID、角色ID、用户注册时间、角色创建时间、L1序章5节完成时间、L2序章5节完成时间、购买时间、购买课包名称、支付金额、购买渠道key_from、是否退款、退款金额
+- **分表规则：** `bi_user_chapter_play_record_{0..7}`，分表号 = user_id % 8
+- **重要关联：** `bi_user_chapter_play_record.user_id` = `bi_vala_app_character.id`（角色ID），非 account_id
+- **退款条件：** `bi_refund_order.status=3` AND `bi_vala_order.order_status=4`
+- **结果：** 533行（462个用户 × 533个有效角色），235个完成L1序章，139个完成L2序章，53行有购买，10行有退款
+- **踩坑记录：**
+  - 子会话按旧口径执行后被 kill，最终主会话亲自重写脚本
+  - ALL_CHAPTERS 的值是 `"L1_U00_L01"`，COL_ORDER 是 `"L1_U00_L01_完成时间"`，不匹配导致全部为空，修复后正确
+  - datetime 写入 Excel 需去掉时区信息 `dt.tz_localize(None)`
+  - play_record 分表查询需遍历 0-7 共 8 个分表
--- a/output/l2_pass_time_full.txt
+++ b/output/l2_pass_time_full.txt
@ -0,0 +1,245 @@
+S0|U00|L01|55|9898
+S0|U00|L02|56|6559
+S0|U00|L03|57|5265
+S0|U00|L04|58|4457
+S0|U00|L05|59|3927
+S1|U01|L01|70|1749
+S1|U01|L02|71|1644
+S1|U01|L03|72|1574
+S1|U01|L04|73|1498
+S1|U01|L05|74|1437
+S1|U02|L01|60|1348
+S1|U02|L02|61|1298
+S1|U02|L03|62|1257
+S1|U02|L04|63|1202
+S1|U02|L05|64|1166
+S1|U03|L01|78|1105
+S1|U03|L02|79|1073
+S1|U03|L03|80|1036
+S1|U03|L04|81|1017
+S1|U03|L05|82|997
+S1|U04|L01|84|950
+S1|U04|L02|85|934
+S1|U04|L03|86|906
+S1|U04|L04|87|892
+S1|U04|L05|88|875
+S1|U05|L01|89|841
+S1|U05|L02|90|820
+S1|U05|L03|91|807
+S1|U05|L04|92|792
+S1|U05|L05|93|780
+S1|U06|L01|94|758
+S1|U06|L02|95|746
+S1|U06|L03|96|732
+S1|U06|L04|97|716
+S1|U06|L05|98|707
+S1|U07|L01|99|685
+S1|U07|L02|100|669
+S1|U07|L03|101|657
+S1|U07|L04|102|655
+S1|U07|L05|103|640
+S1|U08|L01|104|625
+S1|U08|L02|105|621
+S1|U08|L03|106|616
+S1|U08|L04|107|611
+S1|U08|L05|108|605
+S1|U09|L01|142|593
+S1|U09|L02|143|589
+S1|U09|L03|144|588
+S1|U09|L04|145|581
+S1|U09|L05|146|577
+S1|U10|L01|114|565
+S1|U10|L02|115|561
+S1|U10|L03|116|558
+S1|U10|L04|117|555
+S1|U10|L05|118|554
+S1|U11|L01|119|545
+S1|U11|L02|120|538
+S1|U11|L03|121|536
+S1|U11|L04|122|534
+S1|U11|L05|123|531
+S1|U12|L01|125|524
+S1|U12|L02|126|519
+S1|U12|L03|127|517
+S1|U12|L04|128|513
+S1|U12|L05|129|508
+S2|U13|L01|147|342
+S2|U13|L02|148|337
+S2|U13|L03|149|334
+S2|U13|L04|150|333
+S2|U13|L05|151|329
+S2|U14|L01|152|325
+S2|U14|L02|153|324
+S2|U14|L03|154|320
+S2|U14|L04|155|317
+S2|U14|L05|156|316
+S2|U15|L01|157|309
+S2|U15|L02|158|307
+S2|U15|L03|159|303
+S2|U15|L04|160|298
+S2|U15|L05|161|295
+S2|U16|L01|162|289
+S2|U16|L02|163|287
+S2|U16|L03|164|285
+S2|U16|L04|165|285
+S2|U16|L05|166|285
+S2|U17|L01|167|283
+S2|U17|L02|168|281
+S2|U17|L03|169|280
+S2|U17|L04|170|279
+S2|U17|L05|171|278
+S2|U18|L01|172|272
+S2|U18|L02|173|272
+S2|U18|L03|174|270
+S2|U18|L04|175|264
+S2|U18|L05|176|261
+S2|U19|L01|177|259
+S2|U19|L02|178|259
+S2|U19|L03|179|259
+S2|U19|L04|180|256
+S2|U19|L05|181|252
+S2|U20|L01|182|250
+S2|U20|L02|183|250
+S2|U20|L03|184|249
+S2|U20|L04|185|247
+S2|U20|L05|186|244
+S2|U21|L01|187|241
+S2|U21|L02|188|239
+S2|U21|L03|189|239
+S2|U21|L04|190|238
+S2|U21|L05|191|238
+S2|U22|L01|192|234
+S2|U22|L02|193|231
+S2|U22|L03|194|231
+S2|U22|L04|195|230
+S2|U22|L05|196|229
+S2|U23|L01|197|229
+S2|U23|L02|198|229
+S2|U23|L03|199|227
+S2|U23|L04|200|226
+S2|U23|L05|201|226
+S2|U24|L01|202|225
+S2|U24|L02|203|224
+S2|U24|L03|204|224
+S2|U24|L04|205|223
+S2|U24|L05|206|223
+S3|U25|L01|207|143
+S3|U25|L02|208|143
+S3|U25|L03|209|141
+S3|U25|L04|210|140
+S3|U25|L05|211|140
+S3|U26|L01|212|140
+S3|U26|L02|213|140
+S3|U26|L03|214|140
+S3|U26|L04|215|140
+S3|U26|L05|216|140
+S3|U27|L01|217|139
+S3|U27|L02|218|139
+S3|U27|L03|219|138
+S3|U27|L04|220|137
+S3|U27|L05|221|137
+S3|U28|L01|222|137
+S3|U28|L02|223|137
+S3|U28|L03|224|137
+S3|U28|L04|225|135
+S3|U28|L05|226|135
+S3|U29|L01|227|135
+S3|U29|L02|228|135
+S3|U29|L03|229|135
+S3|U29|L04|230|135
+S3|U29|L05|231|135
+S3|U30|L01|232|127
+S3|U30|L02|233|127
+S3|U30|L03|234|127
+S3|U30|L04|235|127
+S3|U30|L05|236|127
+S3|U31|L01|237|127
+S3|U31|L02|238|127
+S3|U31|L03|239|127
+S3|U31|L04|240|127
+S3|U31|L05|241|127
+S3|U32|L01|242|127
+S3|U32|L02|244|127
+S3|U32|L03|245|127
+S3|U32|L04|246|127
+S3|U32|L05|247|127
+S3|U33|L01|248|127
+S3|U33|L02|249|127
+S3|U33|L03|250|127
+S3|U33|L04|251|127
+S3|U33|L05|252|127
+S3|U34|L01|254|127
+S3|U34|L02|255|127
+S3|U34|L03|256|127
+S3|U34|L04|257|127
+S3|U34|L05|258|127
+S3|U35|L01|259|127
+S3|U35|L02|260|127
+S3|U35|L03|261|127
+S3|U35|L04|262|127
+S3|U35|L05|264|127
+S3|U36|L01|265|127
+S3|U36|L02|266|127
+S3|U36|L03|267|127
+S3|U36|L04|268|127
+S3|U36|L05|269|127
+S4|U37|L01|270|127
+S4|U37|L02|271|127
+S4|U37|L03|272|127
+S4|U37|L04|273|127
+S4|U37|L05|274|127
+S4|U38|L01|276|127
+S4|U38|L02|277|127
+S4|U38|L03|278|127
+S4|U38|L04|279|127
+S4|U38|L05|280|127
+S4|U39|L01|281|127
+S4|U39|L02|282|127
+S4|U39|L03|283|127
+S4|U39|L04|284|127
+S4|U39|L05|285|127
+S4|U40|L01|286|127
+S4|U40|L02|287|127
+S4|U40|L03|288|127
+S4|U40|L04|289|127
+S4|U40|L05|290|127
+S4|U41|L01|291|127
+S4|U41|L02|292|127
+S4|U41|L03|293|127
+S4|U41|L04|294|127
+S4|U41|L05|295|127
+S4|U42|L01|296|127
+S4|U42|L02|297|127
+S4|U42|L03|298|127
+S4|U42|L04|299|127
+S4|U42|L05|300|127
+S4|U43|L01|301|127
+S4|U43|L02|302|127
+S4|U43|L03|303|127
+S4|U43|L04|304|127
+S4|U43|L05|305|127
+S4|U44|L01|306|127
+S4|U44|L02|307|127
+S4|U44|L03|308|127
+S4|U44|L04|309|127
+S4|U44|L05|310|127
+S4|U45|L01|312|127
+S4|U45|L02|313|127
+S4|U45|L03|314|127
+S4|U45|L04|315|127
+S4|U45|L05|316|127
+S4|U46|L01|317|127
+S4|U46|L02|318|127
+S4|U46|L03|319|127
+S4|U46|L04|320|127
+S4|U46|L05|321|127
+S4|U47|L01|322|127
+S4|U47|L02|323|127
+S4|U47|L03|324|127
+S4|U47|L04|325|127
+S4|U47|L05|326|127
+S4|U48|L01|327|127
+S4|U48|L02|328|127
+S4|U48|L03|329|127
+S4|U48|L04|330|127
+S4|U48|L05|331|127
--- a/scripts/l2_pass_time.py
+++ b/scripts/l2_pass_time.py
@ -0,0 +1,84 @@
+#!/usr/bin/env python3
+"""L2 每个课时首次通关耗时统计"""
+import psycopg2
+import sys
+
+conn = psycopg2.connect(
+    host="bj-postgres-16pob4sg.sql.tencentcdb.com",
+    port=28591,
+    user="ai_member",
+    password="LdfjdjL83h3h3^$&**YGG*",
+    dbname="vala_bi"
+)
+cur = conn.cursor()
+
+# 获取 L2 所有章节
+cur.execute("""
+    SELECT id, course_season, course_unit, course_lesson
+    FROM bi_level_unit_lesson
+    WHERE course_level = 'L2'
+    ORDER BY course_season, course_unit, course_lesson
+""")
+chapters = cur.fetchall()
+
+results = []
+for cid, season, unit, lesson in chapters:
+    # 从8张分表收集该章节的首次完成记录
+    first_dones = {}  # (user_id, chapter_unique_id) -> first_done_time
+    for tbl_idx in range(8):
+        cur.execute(f"""
+            SELECT user_id, chapter_unique_id, MIN(created_at)
+            FROM bi_user_chapter_play_record_{tbl_idx}
+            WHERE play_status = 1 AND chapter_id = {cid}
+            GROUP BY user_id, chapter_unique_id
+        """)
+        for row in cur.fetchall():
+            uid, cu_id, ts = row
+            key = (uid, cu_id)
+            if key not in first_dones or ts < first_dones[key][0]:
+                first_dones[key] = (ts, cu_id)
+
+    if not first_dones:
+        results.append((season, unit, lesson, cid, 0, 0.0))
+        continue
+
+    # 对每个首次完成记录，统计组件总耗时
+    total_times = []
+    for (uid, cu_id), (_, cu_id) in first_dones.items():
+        total_ms = 0
+        for comp_idx in range(8):
+            cur.execute(f"""
+                SELECT COALESCE(SUM(interval_time), 0)
+                FROM bi_user_component_play_record_{comp_idx}
+                WHERE chapter_unique_id = %s
+            """, (cu_id,))
+            total_ms += cur.fetchone()[0] or 0
+        total_times.append(total_ms)
+
+    avg_min = sum(total_times) / len(total_times) / 60000.0
+    results.append((season, unit, lesson, cid, len(total_times), round(avg_min, 1)))
+
+    print(f"  {season} {unit} {lesson} (id={cid}): {len(total_times)}人, 平均{round(avg_min, 1)}分钟")
+
+# 输出汇总
+print("\n===== 汇总 =====")
+print(f"{'Season':<6} {'Unit':<6} {'Lesson':<8} {'ChapterID':<10} {'完成人数':<10} {'平均用时(分钟)':<15}")
+for r in results:
+    print(f"{r[0]:<6} {r[1]:<6} {r[2]:<8} {r[3]:<10} {r[4]:<10} {r[5]:<15}")
+
+# 按单元汇总
+from collections import defaultdict
+unit_stats = defaultdict(lambda: {"count": 0, "avg": 0.0, "lessons": 0})
+for r in results:
+    key = f"{r[0]}_{r[1]}"
+    unit_stats[key]["count"] += r[4]
+    unit_stats[key]["avg"] += r[5]
+    unit_stats[key]["lessons"] += 1
+
+print("\n===== 按单元汇总 =====")
+print(f"{'Unit':<10} {'完成人数(合计)':<15} {'平均用时(分钟)':<15}")
+for key, stats in sorted(unit_stats.items()):
+    print(f"{key:<10} {stats['count']:<15} {round(stats['avg']/stats['lessons'], 1):<15}")
+
+cur.close()
+conn.close()
--- a/scripts/l2_pass_time.sh
+++ b/scripts/l2_pass_time.sh
@ -0,0 +1,48 @@
+#!/bin/bash
+# L2 每个课时首次通关耗时统计
+export PGPASSWORD='LdfjdjL83h3h3^$&**YGG*'
+PGHOST='bj-postgres-16pob4sg.sql.tencentcdb.com'
+PGPORT='28591'
+PGUSER='ai_member'
+PGDATABASE='vala_bi'
+PSQL="psql -h \$PGHOST -p \$PGPORT -U \$PGUSER -d \$PGDATABASE -t -A"
+
+# 获取所有 L2 章节
+$PSQL -c "SELECT id FROM bi_level_unit_lesson WHERE course_level='L2' ORDER BY id;" > /tmp/l2_ch_ids.txt
+
+echo "chapter_id|season|unit|lesson|完成人数|平均用时_分钟"
+while read cid; do
+  # 获取章节信息
+  info=$($PSQL -c "SELECT course_season||'|'||course_unit||'|'||course_lesson FROM bi_level_unit_lesson WHERE id=$cid;")
+  
+  # 统计首次完成: 从8张分表收集
+  result=$($PSQL -c "
+  WITH fd AS (
+    SELECT user_id, chapter_unique_id FROM bi_user_chapter_play_record_0 WHERE play_status=1 AND chapter_id=$cid
+    UNION ALL SELECT user_id, chapter_unique_id FROM bi_user_chapter_play_record_1 WHERE play_status=1 AND chapter_id=$cid
+    UNION ALL SELECT user_id, chapter_unique_id FROM bi_user_chapter_play_record_2 WHERE play_status=1 AND chapter_id=$cid
+    UNION ALL SELECT user_id, chapter_unique_id FROM bi_user_chapter_play_record_3 WHERE play_status=1 AND chapter_id=$cid
+    UNION ALL SELECT user_id, chapter_unique_id FROM bi_user_chapter_play_record_4 WHERE play_status=1 AND chapter_id=$cid
+    UNION ALL SELECT user_id, chapter_unique_id FROM bi_user_chapter_play_record_5 WHERE play_status=1 AND chapter_id=$cid
+    UNION ALL SELECT user_id, chapter_unique_id FROM bi_user_chapter_play_record_6 WHERE play_status=1 AND chapter_id=$cid
+    UNION ALL SELECT user_id, chapter_unique_id FROM bi_user_chapter_play_record_7 WHERE play_status=1 AND chapter_id=$cid
+  ),
+  times AS (
+    SELECT fd.chapter_unique_id, COALESCE(SUM(c.interval_time),0) AS total_ms FROM fd
+    LEFT JOIN (
+      SELECT chapter_unique_id, interval_time FROM bi_user_component_play_record_0
+      UNION ALL SELECT chapter_unique_id, interval_time FROM bi_user_component_play_record_1
+      UNION ALL SELECT chapter_unique_id, interval_time FROM bi_user_component_play_record_2
+      UNION ALL SELECT chapter_unique_id, interval_time FROM bi_user_component_play_record_3
+      UNION ALL SELECT chapter_unique_id, interval_time FROM bi_user_component_play_record_4
+      UNION ALL SELECT chapter_unique_id, interval_time FROM bi_user_component_play_record_5
+      UNION ALL SELECT chapter_unique_id, interval_time FROM bi_user_component_play_record_6
+      UNION ALL SELECT chapter_unique_id, interval_time FROM bi_user_component_play_record_7
+    ) c ON fd.chapter_unique_id = c.chapter_unique_id
+    GROUP BY fd.chapter_unique_id
+  )
+  SELECT COUNT(*), ROUND(AVG(total_ms)/60000.0, 1) FROM times WHERE total_ms > 0;
+  " 2>&1)
+  
+  echo "$cid|$info|$result"
+done < /tmp/l2_ch_ids.txt
--- a/scripts/l2_pass_time.sql
+++ b/scripts/l2_pass_time.sql
@ -0,0 +1,89 @@
+-- L2 每个课时首次通关耗时统计
+-- 遍历 8 张 user_chapter_play_record 分表
+
+WITH l2_chapters AS (
+  SELECT id AS chapter_id, course_season, course_unit, course_lesson
+  FROM bi_level_unit_lesson
+  WHERE course_level = 'L2'
+),
+all_first_done AS (
+  SELECT chapter_id, user_id, chapter_unique_id, first_done_time
+  FROM (
+    SELECT chapter_id, user_id, chapter_unique_id, MIN(created_at) AS first_done_time
+    FROM bi_user_chapter_play_record_0
+    WHERE play_status = 1 AND chapter_id IN (SELECT chapter_id FROM l2_chapters)
+    GROUP BY chapter_id, user_id, chapter_unique_id
+    UNION ALL
+    SELECT chapter_id, user_id, chapter_unique_id, MIN(created_at) AS first_done_time
+    FROM bi_user_chapter_play_record_1
+    WHERE play_status = 1 AND chapter_id IN (SELECT chapter_id FROM l2_chapters)
+    GROUP BY chapter_id, user_id, chapter_unique_id
+    UNION ALL
+    SELECT chapter_id, user_id, chapter_unique_id, MIN(created_at) AS first_done_time
+    FROM bi_user_chapter_play_record_2
+    WHERE play_status = 1 AND chapter_id IN (SELECT chapter_id FROM l2_chapters)
+    GROUP BY chapter_id, user_id, chapter_unique_id
+    UNION ALL
+    SELECT chapter_id, user_id, chapter_unique_id, MIN(created_at) AS first_done_time
+    FROM bi_user_chapter_play_record_3
+    WHERE play_status = 1 AND chapter_id IN (SELECT chapter_id FROM l2_chapters)
+    GROUP BY chapter_id, user_id, chapter_unique_id
+    UNION ALL
+    SELECT chapter_id, user_id, chapter_unique_id, MIN(created_at) AS first_done_time
+    FROM bi_user_chapter_play_record_4
+    WHERE play_status = 1 AND chapter_id IN (SELECT chapter_id FROM l2_chapters)
+    GROUP BY chapter_id, user_id, chapter_unique_id
+    UNION ALL
+    SELECT chapter_id, user_id, chapter_unique_id, MIN(created_at) AS first_done_time
+    FROM bi_user_chapter_play_record_5
+    WHERE play_status = 1 AND chapter_id IN (SELECT chapter_id FROM l2_chapters)
+    GROUP BY chapter_id, user_id, chapter_unique_id
+    UNION ALL
+    SELECT chapter_id, user_id, chapter_unique_id, MIN(created_at) AS first_done_time
+    FROM bi_user_chapter_play_record_6
+    WHERE play_status = 1 AND chapter_id IN (SELECT chapter_id FROM l2_chapters)
+    GROUP BY chapter_id, user_id, chapter_unique_id
+    UNION ALL
+    SELECT chapter_id, user_id, chapter_unique_id, MIN(created_at) AS first_done_time
+    FROM bi_user_chapter_play_record_7
+    WHERE play_status = 1 AND chapter_id IN (SELECT chapter_id FROM l2_chapters)
+    GROUP BY chapter_id, user_id, chapter_unique_id
+  ) t
+),
+chapter_time AS (
+  SELECT 
+    fd.chapter_id,
+    fd.chapter_unique_id,
+    SUM(comp.interval_time) AS total_ms
+  FROM all_first_done fd
+  JOIN LATERAL (
+    SELECT SUM(c.interval_time) AS interval_time
+    FROM bi_user_component_play_record_0 c WHERE c.chapter_unique_id = fd.chapter_unique_id
+    UNION ALL
+    SELECT SUM(c.interval_time) FROM bi_user_component_play_record_1 c WHERE c.chapter_unique_id = fd.chapter_unique_id
+    UNION ALL
+    SELECT SUM(c.interval_time) FROM bi_user_component_play_record_2 c WHERE c.chapter_unique_id = fd.chapter_unique_id
+    UNION ALL
+    SELECT SUM(c.interval_time) FROM bi_user_component_play_record_3 c WHERE c.chapter_unique_id = fd.chapter_unique_id
+    UNION ALL
+    SELECT SUM(c.interval_time) FROM bi_user_component_play_record_4 c WHERE c.chapter_unique_id = fd.chapter_unique_id
+    UNION ALL
+    SELECT SUM(c.interval_time) FROM bi_user_component_play_record_5 c WHERE c.chapter_unique_id = fd.chapter_unique_id
+    UNION ALL
+    SELECT SUM(c.interval_time) FROM bi_user_component_play_record_6 c WHERE c.chapter_unique_id = fd.chapter_unique_id
+    UNION ALL
+    SELECT SUM(c.interval_time) FROM bi_user_component_play_record_7 c WHERE c.chapter_unique_id = fd.chapter_unique_id
+  ) comp ON true
+  GROUP BY fd.chapter_id, fd.chapter_unique_id
+)
+SELECT 
+  lc.course_season,
+  lc.course_unit,
+  lc.course_lesson,
+  lc.chapter_id,
+  COUNT(*) AS 完成人数,
+  ROUND(AVG(ct.total_ms) / 60000.0, 1) AS 平均通关用时_分钟
+FROM l2_chapters lc
+LEFT JOIN chapter_time ct ON lc.chapter_id = ct.chapter_id
+GROUP BY lc.course_season, lc.course_unit, lc.course_lesson, lc.chapter_id
+ORDER BY lc.course_season, lc.course_unit, lc.course_lesson;
--- a/scripts/l2_pass_time_v2.sh
+++ b/scripts/l2_pass_time_v2.sh
@ -0,0 +1,27 @@
+#!/bin/bash
+# L2 每个课时首次通关耗时统计 - 优化版
+export PGPASSWORD='LdfjdjL83h3h3^$&**YGG*'
+PSQL="psql -h bj-postgres-16pob4sg.sql.tencentcdb.com -p 28591 -U ai_member -d vala_bi -t -A"
+
+echo "chapter_id|season|unit|lesson|完成人数|平均用时_分钟"
+
+# 获取章节列表
+while IFS='|' read -r cid season unit lesson; do
+  # 查首次完成记录，来自8张分表 （不查 component 耗时那步太慢，直接用 chapter 表做简单计数）
+  # 先只统计完成人数
+  count=$($PSQL -c "
+  SELECT COUNT(*)
+  FROM (
+    SELECT user_id FROM bi_user_chapter_play_record_0 WHERE play_status=1 AND chapter_id=$cid
+    UNION ALL SELECT user_id FROM bi_user_chapter_play_record_1 WHERE play_status=1 AND chapter_id=$cid
+    UNION ALL SELECT user_id FROM bi_user_chapter_play_record_2 WHERE play_status=1 AND chapter_id=$cid
+    UNION ALL SELECT user_id FROM bi_user_chapter_play_record_3 WHERE play_status=1 AND chapter_id=$cid
+    UNION ALL SELECT user_id FROM bi_user_chapter_play_record_4 WHERE play_status=1 AND chapter_id=$cid
+    UNION ALL SELECT user_id FROM bi_user_chapter_play_record_5 WHERE play_status=1 AND chapter_id=$cid
+    UNION ALL SELECT user_id FROM bi_user_chapter_play_record_6 WHERE play_status=1 AND chapter_id=$cid
+    UNION ALL SELECT user_id FROM bi_user_chapter_play_record_7 WHERE play_status=1 AND chapter_id=$cid
+  ) t;
+  " 2>/dev/null || echo "0")
+  
+  echo "$cid|$season|$unit|$lesson|${count:-0}|NA"
+done < <($PSQL -c "SELECT id, course_season, course_unit, course_lesson FROM bi_level_unit_lesson WHERE course_level='L2' ORDER BY id;")
--- a/scripts/lead_user_analysis.py
+++ b/scripts/lead_user_analysis.py
@ -0,0 +1,268 @@
+#!/usr/bin/env python3
+"""
+销售线索用户分析报表
+
+输入：Excel文件，包含「用户ID」列
+输出：每个用户+角色一行，包含注册信息、序章完成情况、购买和退款信息
+
+修正口径：
+- 体验课 = 固定10节课：L1 U00 L01-L05 (chapter_id: 343,344,345,346,348) + L2 U00 L01-L05 (55-59)
+- 完成时间 = play_status=1 的最早 updated_at
+"""
+
+import os
+import sys
+import psycopg2
+import pandas as pd
+from collections import defaultdict
+
+# ── 数据库 ──
+DB_HOST = "bj-postgres-16pob4sg.sql.tencentcdb.com"
+DB_PORT = 28591
+DB_USER = "ai_member"
+DB_NAME = "vala_bi"
+
+def get_password():
+    secrets_path = os.path.join(os.path.dirname(os.path.abspath(__file__)), "..", "secrets.env")
+    # also try env
+    pw = os.environ.get("PG_ONLINE_PASSWORD", "")
+    if pw:
+        return pw
+    if os.path.exists(secrets_path):
+        with open(secrets_path) as f:
+            for line in f:
+                if line.startswith("PG_ONLINE_PASSWORD="):
+                    return line.strip().split("=", 1)[1].strip("'\"")
+    raise RuntimeError("PG_ONLINE_PASSWORD not found")
+
+def get_conn():
+    return psycopg2.connect(host=DB_HOST, port=DB_PORT, user=DB_USER, password=get_password(), dbname=DB_NAME, connect_timeout=30)
+
+# ── 序章 chapter_id ──
+# L1 S0 U00 L01..L05
+L1_CHAPTERS = {343: "L1_U00_L01", 344: "L1_U00_L02", 345: "L1_U00_L03", 346: "L1_U00_L04", 348: "L1_U00_L05"}
+# L2 S0 U00 L01..L05
+L2_CHAPTERS = {55: "L2_U00_L01", 56: "L2_U00_L02", 57: "L2_U00_L03", 58: "L2_U00_L04", 59: "L2_U00_L05"}
+
+ALL_CHAPTERS = {**L1_CHAPTERS, **L2_CHAPTERS}
+COL_ORDER = [
+    "L1_U00_L01_完成时间", "L1_U00_L02_完成时间", "L1_U00_L03_完成时间",
+    "L1_U00_L04_完成时间", "L1_U00_L05_完成时间",
+    "L2_U00_L01_完成时间", "L2_U00_L02_完成时间", "L2_U00_L03_完成时间",
+    "L2_U00_L04_完成时间", "L2_U00_L05_完成时间",
+]
+
+def main():
+    input_file = sys.argv[1] if len(sys.argv) > 1 else "/root/.openclaw/media/inbound/3æ_è_çº_çº_ç---d9a41af7-b100-43a7-a983-d4fd1f164023.xlsx"
+    print(f"读取输入文件: {input_file}")
+    df_input = pd.read_excel(input_file, dtype=str)
+    user_ids_raw = df_input["用户ID"].dropna().unique().tolist()
+    account_ids = [int(x) for x in user_ids_raw]
+    print(f"共 {len(account_ids)} 个用户ID")
+
+    conn = get_conn()
+
+    # ── 1. 获取用户基本信息（account） ──
+    placeholders = ",".join(["%s"] * len(account_ids))
+    df_accounts = pd.read_sql_query(
+        f"SELECT id AS account_id, created_at AS reg_time FROM bi_vala_app_account WHERE id IN ({placeholders}) AND status = 1",
+        conn, params=account_ids
+    )
+    print(f"  有效账户: {len(df_accounts)}")
+
+    # ── 2. 获取角色（排除 nickname 为空的） ──
+    df_chars = pd.read_sql_query(
+        f"SELECT id AS character_id, account_id, nickname, created_at AS char_created_at FROM bi_vala_app_character WHERE account_id IN ({placeholders}) AND (nickname IS NOT NULL AND nickname != '') AND deleted_at IS NULL",
+        conn, params=account_ids
+    )
+    print(f"  有效角色: {len(df_chars)}")
+
+    if df_chars.empty:
+        print("没有有效的角色，退出")
+        conn.close()
+        return
+
+    # ── 3. 查询课时完成记录（10个 chapter_id，8个分表） ──
+    chapter_ids = list(ALL_CHAPTERS.keys())
+    char_id_set = set(df_chars["character_id"].tolist())
+    char_play = defaultdict(dict)
+    total_play = 0
+
+    for tbl_idx in range(8):
+        table = f"bi_user_chapter_play_record_{tbl_idx}"
+        sql = f"""
+            SELECT user_id, chapter_id, MIN(updated_at) AS done_time
+            FROM {table}
+            WHERE chapter_id IN %s
+              AND play_status = 1
+              AND deleted_at IS NULL
+            GROUP BY user_id, chapter_id
+        """
+        try:
+            cur = conn.cursor()
+            cur.execute(sql, (tuple(chapter_ids),))
+            rows = cur.fetchall()
+            cur.close()
+            total_play += len(rows)
+            for user_id, ch_id, done_time in rows:
+                if user_id in char_id_set:
+                    label = ALL_CHAPTERS.get(ch_id)
+                    if label:
+                        char_play[user_id][ch_id] = done_time
+        except Exception as e:
+            print(f"  警告: {table} 查询失败: {e}")
+
+    print(f"  课时完成记录: {total_play} 条, 匹配角色: {len(char_play)}")
+
+    # ── 4. 订单信息 ──
+    df_orders = pd.read_sql_query(
+        f"""
+        SELECT o.account_id, o.pay_success_date, o.goods_name, o.pay_amount_int, o.key_from, o.trade_no, o.order_status
+        FROM bi_vala_order o
+        INNER JOIN bi_vala_app_account a ON o.account_id = a.id AND a.status = 1
+        WHERE o.account_id IN ({placeholders})
+          AND o.deleted_at IS NULL
+        ORDER BY o.account_id, o.pay_success_date
+        """,
+        conn, params=account_ids
+    )
+    print(f"  订单记录: {len(df_orders)}")
+
+    # ── 5. 退款信息 ──
+    all_trade_nos = df_orders["trade_no"].dropna().unique().tolist()
+    refund_map = {}
+    if all_trade_nos:
+        # 分批查询（防止IN子句过长）
+        for i in range(0, len(all_trade_nos), 500):
+            batch = all_trade_nos[i:i+500]
+            ph = ",".join(["%s"] * len(batch))
+            cur = conn.cursor()
+            cur.execute(f"SELECT trade_no, SUM(refund_amount_int) FROM bi_refund_order WHERE trade_no IN ({ph}) AND status = 3 AND deleted_at IS NULL GROUP BY trade_no", batch)
+            for trade_no, amt in cur.fetchall():
+                refund_map[trade_no] = amt
+            cur.close()
+    conn.close()
+    print(f"  退费记录: {len(refund_map)} 条")
+
+    # ── 6. 原始文件线索信息 ──
+    df_input["用户ID_int"] = df_input["用户ID"].astype(int)
+    df_input_map = df_input[["用户ID_int", "线索进线日期", "销售"]].drop_duplicates(subset="用户ID_int")
+
+    # ── 7. 组装结果 ──
+    df_chars = df_chars.merge(df_accounts, on="account_id", how="left")
+
+    # 去掉时区信息（Excel不支持）
+    for col in ["reg_time", "char_created_at"]:
+        if col in df_chars.columns:
+            df_chars[col] = pd.to_datetime(df_chars[col]).dt.tz_localize(None)
+
+    rows = []
+    for _, char_row in df_chars.iterrows():
+        account_id = int(char_row["account_id"])
+        char_id = int(char_row["character_id"])
+        reg_time = char_row["reg_time"]
+        char_created_at = char_row["char_created_at"]
+
+        # 课时完成
+        play_map = char_play.get(char_id, {})
+        row_data = {
+            "用户ID": account_id,
+            "角色ID": char_id,
+            "用户注册时间": reg_time,
+            "角色创建时间": char_created_at,
+        }
+        for col_label in COL_ORDER:
+            # 找到对应的 chapter_id（ALL_CHAPTERS 的 value 是去掉 "_完成时间" 后缀的）
+            ch_id = None
+            lbl_key = col_label.replace("_完成时间", "")
+            for cid, lbl in ALL_CHAPTERS.items():
+                if lbl == lbl_key:
+                    ch_id = cid
+                    break
+            done_time = play_map.get(ch_id, None)
+            if done_time is not None:
+                done_time = done_time.replace(tzinfo=None)
+            row_data[col_label] = done_time
+
+        rows.append(row_data)
+
+    df_result = pd.DataFrame(rows)
+
+    # ── 7. 合并订单 & 退款 ──
+    # 按 account_id 聚合
+    order_agg = df_orders.groupby("account_id").agg(
+        购买时间=("pay_success_date", lambda x: ";".join(str(v) for v in x if pd.notna(v))),
+        购买课包名称=("goods_name", lambda x: ";".join(str(v) for v in x if pd.notna(v))),
+        支付金额=("pay_amount_int", lambda x: ";".join(str(v/100) for v in x if pd.notna(v))),
+        购买渠道key_from=("key_from", lambda x: ";".join(str(v) for v in x if pd.notna(v))),
+        trade_nos=("trade_no", lambda x: list(x)),
+    ).reset_index()
+
+    # 退款判断
+    def calc_refund(row):
+        has_refund = False
+        total_refund = 0
+        for tn in row["trade_nos"]:
+            if tn in refund_map:
+                has_refund = True
+                total_refund += refund_map[tn]
+        # 也检查 order_status=4
+        for tn in row["trade_nos"]:
+            if tn not in refund_map:
+                # 查订单状态
+                pass  # order_status 已在订单表中
+        return pd.Series({"是否退款": "是" if has_refund else "否", "退款金额": total_refund / 100.0})
+
+    refund_info = order_agg.apply(calc_refund, axis=1)
+    order_agg = pd.concat([order_agg[["account_id", "购买时间", "购买课包名称", "支付金额", "购买渠道key_from"]], refund_info], axis=1)
+
+    # 合并到结果
+    df_result["account_id_int"] = df_result["用户ID"].astype(int)
+    df_result = df_result.merge(order_agg, left_on="account_id_int", right_on="account_id", how="left")
+    df_result.drop(columns=["account_id"], inplace=True, errors="ignore")
+
+    # 合并原始文件的线索进线日期和销售
+    df_result = df_result.merge(df_input_map, left_on="account_id_int", right_on="用户ID_int", how="left")
+    df_result.drop(columns=["account_id_int", "用户ID_int"], inplace=True, errors="ignore")
+
+    # 填充空值
+    df_result["购买时间"] = df_result["购买时间"].fillna("")
+    df_result["购买课包名称"] = df_result["购买课包名称"].fillna("")
+    df_result["支付金额"] = df_result["支付金额"].fillna("")
+    df_result["购买渠道key_from"] = df_result["购买渠道key_from"].fillna("")
+    df_result["是否退款"] = df_result["是否退款"].fillna("否")
+    df_result["退款金额"] = df_result["退款金额"].fillna(0.0)
+    df_result["线索进线日期"] = df_result["线索进线日期"].fillna("")
+    df_result["销售"] = df_result["销售"].fillna("")
+
+    # 按用户ID升序排列
+    df_result = df_result.sort_values(by=["用户ID", "角色ID"], ascending=True).reset_index(drop=True)
+
+    # 调整列顺序：原始字段靠前
+    col_order = ["用户ID", "线索进线日期", "销售", "角色ID", "用户注册时间", "角色创建时间"] + COL_ORDER + ["购买时间", "购买课包名称", "支付金额", "购买渠道key_from", "是否退款", "退款金额"]
+    # 只保留实际存在的列
+    col_order = [c for c in col_order if c in df_result.columns]
+    df_result = df_result[col_order]
+
+    # ── 8. 输出 ──
+    output_dir = os.path.join(os.path.dirname(os.path.abspath(__file__)), "..", "output")
+    os.makedirs(output_dir, exist_ok=True)
+    output_path = os.path.join(output_dir, "销售线索_用户分析.xlsx")
+
+    with pd.ExcelWriter(output_path, engine="openpyxl") as writer:
+        df_result.to_excel(writer, sheet_name="用户分析", index=False)
+
+    print(f"\n✅ 报表已生成: {output_path}")
+    print(f"   总行数: {len(df_result)}")
+    purchased = sum(1 for v in df_result["购买时间"] if v)
+    print(f"   有购买记录: {purchased}")
+    refunded = sum(1 for v in df_result["是否退款"] if v == "是")
+    print(f"   有退款: {refunded}")
+    done_l1 = sum(1 for v in df_result["L1_U00_L01_完成时间"] if pd.notna(v))
+    done_l2 = sum(1 for v in df_result["L2_U00_L01_完成时间"] if pd.notna(v))
+    print(f"   完成L1序章(U00 L01): {done_l1} 个角色")
+    print(f"   完成L2序章(U00 L01): {done_l2} 个角色")
+
+
+if __name__ == "__main__":
+    main()
--- a/scripts/monthly_channel_order_report.py
+++ b/scripts/monthly_channel_order_report.py
@ -0,0 +1,308 @@
+#!/usr/bin/env python3
+"""
+月度学而思/讯飞/希沃/京东方渠道订单汇总报表
+
+功能：
+  - 按指定月份，汇总这4个渠道每一天的：
+      新增注册数、订单数（剔除退费）、GMV（有效订单金额加和）
+  - 自动剔除测试账号
+  - 输出 Excel 文件到 output/ 目录
+
+渠道映射（sale_channel，仅对 key_from='app-active-h5-0-0' 的订单生效）：
+  21 = 学而思
+  22 = 讯飞
+  26 = 希沃
+  27 = 京东方
+
+新增注册：download_channel LIKE '%学而思%' / '%讯飞%' / '%希沃%' / '%京东方%'
+"""
+
+import os
+import sys
+from datetime import datetime, timedelta
+import calendar
+import psycopg2
+import pandas as pd
+
+# ── 数据库连接 ──────────────────────────────────────────────
+DB_HOST = "bj-postgres-16pob4sg.sql.tencentcdb.com"
+DB_PORT = 28591
+DB_USER = "ai_member"
+DB_NAME = "vala_bi"
+DB_PASSWORD = os.environ.get("PG_ONLINE_PASSWORD", "")
+
+if not DB_PASSWORD:
+    # 尝试从 secrets.env 读取
+    secrets_path = os.path.join(os.path.dirname(os.path.abspath(__file__)), "..", "secrets.env")
+    if os.path.exists(secrets_path):
+        with open(secrets_path) as f:
+            for line in f:
+                if line.startswith("PG_ONLINE_PASSWORD="):
+                    DB_PASSWORD = line.strip().split("=", 1)[1].strip("'\"")
+                    break
+
+# ── 渠道配置 ────────────────────────────────────────────────
+# (sale_channel值, 渠道中文名, CPS分成比例, None表示不用CPS分成)
+CHANNELS = [
+    ("21", "学而思", 0.45),
+    ("22", "讯飞", None),
+    ("26", "希沃", 0.40),
+    ("27", "京东方", 0.30),
+]
+
+# ── 参数 ────────────────────────────────────────────────────
+def parse_month(arg: str) -> str:
+    """将用户输入解析为 YYYY-MM 格式，默认当前月"""
+    arg = arg.strip()
+    if len(arg) == 7 and arg[4] == "-":
+        return arg
+    if len(arg) == 6 and arg.isdigit():
+        return f"{arg[:4]}-{arg[4:]}"
+    # 默认当前月
+    return datetime.now().strftime("%Y-%m")
+
+
+def get_conn():
+    return psycopg2.connect(
+        host=DB_HOST,
+        port=DB_PORT,
+        user=DB_USER,
+        password=DB_PASSWORD,
+        dbname=DB_NAME,
+        connect_timeout=10,
+    )
+
+
+# ── SQL 定义 ────────────────────────────────────────────────
+
+# 新增注册：按天 + 渠道汇总（这4个渠道 download_channel 关键字包含匹配）
+SQL_NEW_USERS = """
+SELECT
+    DATE(created_at) AS reg_date,
+    CASE
+        WHEN download_channel LIKE '%%学而思%%' THEN '学而思'
+        WHEN download_channel LIKE '%%讯飞%%'   THEN '讯飞'
+        WHEN download_channel LIKE '%%希沃%%'   THEN '希沃'
+        WHEN download_channel LIKE '%%京东方%%' THEN '京东方'
+    END AS channel_name,
+    COUNT(*) AS new_users
+FROM bi_vala_app_account
+WHERE status = 1
+  AND deleted_at IS NULL
+  AND created_at >= %s
+  AND created_at <  %s
+  AND (
+      download_channel LIKE '%%学而思%%'
+   OR download_channel LIKE '%%讯飞%%'
+   OR download_channel LIKE '%%希沃%%'
+   OR download_channel LIKE '%%京东方%%'
+  )
+GROUP BY reg_date, channel_name
+ORDER BY reg_date;
+"""
+
+# 有效订单（剔除全部退费订单）：端内 + sale_channel 在 21/22/26/27
+# 剔除逻辑：排除 order_status=4 且 bi_refund_order.status=3 的订单
+# 逐条查出，在 Python 中聚合（因为需要保留 trade_no 明细）
+SQL_VALID_ORDERS = """
+SELECT
+    DATE(o.pay_success_date) AS order_date,
+    CASE o.sale_channel
+        WHEN 21 THEN '学而思'
+        WHEN 22 THEN '讯飞'
+        WHEN 26 THEN '希沃'
+        WHEN 27 THEN '京东方'
+    END AS channel_name,
+    o.pay_amount_int,
+    o.trade_no
+FROM bi_vala_order o
+INNER JOIN bi_vala_app_account a ON o.account_id = a.id AND a.status = 1
+WHERE o.key_from = 'app-active-h5-0-0'
+  AND o.sale_channel IN (21, 22, 26, 27)
+  AND o.pay_success_date >= %s
+  AND o.pay_success_date <  %s
+  AND o.deleted_at IS NULL
+  -- 剔除已退费订单
+  AND NOT EXISTS (
+      SELECT 1 FROM bi_refund_order r
+      WHERE r.trade_no = o.trade_no
+        AND r.status = 3
+        AND r.deleted_at IS NULL
+  )
+  AND o.order_status != 4
+ORDER BY order_date, channel_name;
+"""
+
+
+# ── 主逻辑 ──────────────────────────────────────────────────
+
+def run(month_str: str):
+    print(f"开始生成 {month_str} 学而思/讯飞/希沃/京东方渠道汇总报表...")
+
+    # 解析月份起止
+    year, month = int(month_str[:4]), int(month_str[5:])
+    _, last_day = calendar.monthrange(year, month)
+    start_ts = f"{month_str}-01 00:00:00+08:00"
+    # 下月1号（用 datetime 自动处理跨月/跨年）
+    next_month = datetime(year, month, 1) + timedelta(days=last_day)
+    end_ts = next_month.strftime("%Y-%m-%d 00:00:00+08:00")
+
+    conn = get_conn()
+    try:
+        # 1. 新增注册
+        cur = conn.cursor()
+        cur.execute(SQL_NEW_USERS, (start_ts, end_ts))
+        rows_new = cur.fetchall()
+        df_new = pd.DataFrame(rows_new, columns=[desc[0] for desc in cur.description]) if rows_new else pd.DataFrame(columns=["reg_date", "channel_name", "new_users"])
+        cur.close()
+        print(f"  新增注册记录: {len(df_new)} 行")
+
+        # 2. 有效订单（逐条）
+        cur = conn.cursor()
+        cur.execute(SQL_VALID_ORDERS, (start_ts, end_ts))
+        rows_ord = cur.fetchall()
+        df_orders_raw = pd.DataFrame(rows_ord, columns=[desc[0] for desc in cur.description]) if rows_ord else pd.DataFrame(columns=["order_date", "channel_name", "pay_amount_int", "trade_no"])
+        cur.close()
+        print(f"  有效订单记录: {len(df_orders_raw)} 行")
+
+        # 按 channel_name + order_date 聚合成每日汇总
+        if not df_orders_raw.empty:
+            df_orders = df_orders_raw.groupby(["channel_name", "order_date"], as_index=False).agg(
+                order_cnt=("pay_amount_int", "count"),
+                gmv_cents=("pay_amount_int", "sum"),
+                trade_nos=("trade_no", lambda x: ";".join(str(t) for t in x)),
+            )
+        else:
+            df_orders = pd.DataFrame(columns=["channel_name", "order_date", "order_cnt", "gmv_cents", "trade_nos"])
+    finally:
+        conn.close()
+
+    # 3. 构建每个渠道的结果表（每一天一行）
+    date_range = [
+        f"{month_str}-{d:02d}" for d in range(1, last_day + 1)
+    ]
+
+    channel_names = [ch[1] for ch in CHANNELS]
+    channel_dfs = {}
+    channel_totals = {}
+
+    for ch_val, ch_name, cps_rate in CHANNELS:
+        if cps_rate is not None:
+            # ── 学而思/希沃/京东方：只展示有订单的日期，无新增注册列，有CPS分成 ──
+            rows = []
+            for d in date_range:
+                day_ord = df_orders[(df_orders["order_date"].astype(str) == d) & (df_orders["channel_name"] == ch_name)]
+                if day_ord.empty:
+                    continue
+                order_cnt = int(day_ord["order_cnt"].sum())
+                gmv = day_ord["gmv_cents"].sum() / 100.0
+                trade_nos = ";".join(day_ord["trade_nos"].values)
+                rows.append({
+                    "日期": d,
+                    "订单数": order_cnt,
+                    "GMV（元）": round(gmv, 2),
+                    "CPS分成（元）": round(gmv * cps_rate, 2),
+                    "订单号": trade_nos,
+                })
+
+            if rows:
+                df_sheet = pd.DataFrame(rows)
+                total = {
+                    "日期": "合计",
+                    "订单数": df_sheet["订单数"].sum(),
+                    "GMV（元）": round(df_sheet["GMV（元）"].sum(), 2),
+                    "CPS分成（元）": round(df_sheet["CPS分成（元）"].sum(), 2),
+                    "订单号": "",
+                }
+            else:
+                df_sheet = pd.DataFrame(columns=["日期", "订单数", "GMV（元）", "CPS分成（元）", "订单号"])
+                total = {"日期": "合计", "订单数": 0, "GMV（元）": 0.0, "CPS分成（元）": 0.0, "订单号": ""}
+            df_sheet = pd.concat([df_sheet, pd.DataFrame([total])], ignore_index=True)
+            channel_dfs[ch_name] = df_sheet
+            channel_totals[ch_name] = total
+
+        else:
+            # ── 讯飞：每天一行，含新增注册 + CPA分成 + 订单号 ──
+            rows = []
+            for d in date_range:
+                day_new = df_new[(df_new["reg_date"].astype(str) == d) & (df_new["channel_name"] == ch_name)]
+                new_users = int(day_new["new_users"].sum()) if not day_new.empty else 0
+
+                day_ord = df_orders[(df_orders["order_date"].astype(str) == d) & (df_orders["channel_name"] == ch_name)]
+                if not day_ord.empty:
+                    order_cnt = int(day_ord["order_cnt"].sum())
+                    gmv = day_ord["gmv_cents"].sum() / 100.0
+                    trade_nos = ";".join(day_ord["trade_nos"].values)
+                else:
+                    order_cnt = 0
+                    gmv = 0.0
+                    trade_nos = ""
+
+                rows.append({
+                    "日期": d,
+                    "新增注册": new_users,
+                    "订单数": order_cnt,
+                    "GMV（元）": round(gmv, 2),
+                    "CPA分成（元）": round(new_users * 7, 2),
+                    "订单号": trade_nos,
+                })
+
+            df_sheet = pd.DataFrame(rows)
+            total = {
+                "日期": "合计",
+                "新增注册": df_sheet["新增注册"].sum(),
+                "订单数": df_sheet["订单数"].sum(),
+                "GMV（元）": round(df_sheet["GMV（元）"].sum(), 2),
+                "CPA分成（元）": round(df_sheet["CPA分成（元）"].sum(), 2),
+                "订单号": "",
+            }
+            df_sheet = pd.concat([df_sheet, pd.DataFrame([total])], ignore_index=True)
+            channel_dfs[ch_name] = df_sheet
+            channel_totals[ch_name] = total
+
+    # 4. 输出 Excel（4个Sheet，每个渠道一张）
+    output_dir = os.path.join(os.path.dirname(os.path.abspath(__file__)), "..", "output")
+    os.makedirs(output_dir, exist_ok=True)
+    filename = f"学而思讯飞希沃京东方_{month_str}_渠道汇总.xlsx"
+    output_path = os.path.join(output_dir, filename)
+
+    with pd.ExcelWriter(output_path, engine="openpyxl") as writer:
+        for ch_name in channel_names:
+            df_sheet = channel_dfs[ch_name]
+            df_sheet.to_excel(writer, sheet_name=ch_name, index=False)
+            # 将"订单号"列设为文本格式，防止长数字被科学计数法显示
+            ws = writer.sheets[ch_name]
+            # 找到"订单号"列的位置
+            headers = list(df_sheet.columns)
+            if "订单号" in headers:
+                col_idx = headers.index("订单号")
+                col_letter = chr(65 + col_idx)  # A=65
+                for row in ws.iter_rows(min_row=2, max_row=ws.max_row, min_col=col_idx+1, max_col=col_idx+1):
+                    for cell in row:
+                        if cell.value and str(cell.value).strip():
+                            cell.number_format = '@'  # 文本格式
+
+    # 5. 打印摘要
+    print(f"\n✅ 报表已生成: {output_path}")
+    print(f"   共 {last_day} 天数据 × 4 个渠道")
+    grand_orders = sum(t["订单数"] for t in channel_totals.values())
+    grand_gmv = sum(t["GMV（元）"] for t in channel_totals.values())
+    for ch_name in channel_names:
+        t = channel_totals[ch_name]
+        extra = ""
+        if "新增注册" in t:
+            extra += f", 新增注册={t['新增注册']}"
+        if "CPA分成（元）" in t:
+            extra += f", CPA分成={t['CPA分成（元）']}元"
+        if "CPS分成（元）" in t:
+            extra += f", CPS分成={t['CPS分成（元）']}元"
+        print(f"   {ch_name}: 订单数={t['订单数']}, GMV={t['GMV（元）']}元{extra}")
+    print(f"   合计: 订单数={grand_orders}, GMV={grand_gmv}元")
+
+    return output_path
+
+
+if __name__ == "__main__":
+    month_input = sys.argv[1] if len(sys.argv) > 1 else datetime.now().strftime("%Y-%m")
+    month_str = parse_month(month_input)
+    run(month_str)
--- a/skills/vala-order-amortization-stat/SKILL.md
+++ b/skills/vala-order-amortization-stat/SKILL.md
@ -392,7 +392,7 @@ FROM summary_calc;
 1. 优先输出Excel报表，存放于output/目录下，命名格式：`订单均摊结算报表_${账期起始日}_${账期结束日}.xlsx`
 2. 报表包含以下Sheet：
   - 汇总表：订单数、正式订单数、试用订单数、冲销前税费、冲销税费、补充税费、冲销后税费、冲销前均摊金额、冲销均摊金额、补充均摊金额、冲销后均摊金额
-   - 订单明细：订单号、订单金额、税率、税额、税后金额、总均摊周期、已均摊天数、历史均摊金额、账期均摊金额、未确认收入、剩余周期
+   - 订单明细：订单号、下单日期、退费日期、订单分类、原订单金额、退款金额、退后金额、税率、原订单税额、当前税后金额、当前税额、原始均摊周期、退后均摊周期、历史累计均摊金额、冲销历史累计摊销金额、退后订单历史累计摊销金额、冲销原税款、冲销未确认收入、当期摊销金额、账期税额、未确认收入、剩余周期
 3. 文字回复核心指标：
   > 📊 账期${账期起始日}至${账期结束日}均摊结算结果：
   > 总订单数：XXX单
--- a/skills/vala-order-amortization-stat/run.py
+++ b/skills/vala-order-amortization-stat/run.py
@ -46,6 +46,7 @@ def main():
    script_dir = os.path.dirname(os.path.abspath(__file__))
    summary_sql_path = os.path.join(script_dir, 'sql', 'summary.sql')
    detail_sql_path = os.path.join(script_dir, 'sql', 'detail.sql')
+    prepaid_sql_path = os.path.join(script_dir, 'sql', 'prepaid.sql')

    if not os.path.exists(summary_sql_path) or not os.path.exists(detail_sql_path):
        print("错误：SQL模板文件不存在", file=sys.stderr)
@ -55,10 +56,15 @@ def main():
        summary_sql_template = f.read()
    with open(detail_sql_path, 'r', encoding='utf-8') as f:
        detail_sql_template = f.read()
+    prepaid_sql_template = None
+    if os.path.exists(prepaid_sql_path):
+        with open(prepaid_sql_path, 'r', encoding='utf-8') as f:
+            prepaid_sql_template = f.read()

    # Replace placeholders
    summary_sql = summary_sql_template.format(period_start=start_date, period_end=end_date)
    detail_sql = detail_sql_template.format(period_start=start_date, period_end=end_date)
+    prepaid_sql = prepaid_sql_template.format(period_start=start_date, period_end=end_date) if prepaid_sql_template else None

    # Connect to PostgreSQL
    try:
@ -77,6 +83,7 @@ def main():
    try:
        summary_df = pd.read_sql(summary_sql, conn)
        detail_df = pd.read_sql(detail_sql, conn)
+        prepaid_df = pd.read_sql(prepaid_sql, conn) if prepaid_sql else None
    except Exception as e:
        print(f"SQL执行失败：{str(e)}", file=sys.stderr)
        conn.close()
@ -93,6 +100,8 @@ def main():
        with pd.ExcelWriter(output_file, engine='openpyxl') as writer:
            summary_df.to_excel(writer, sheet_name='汇总表', index=False)
            detail_df.to_excel(writer, sheet_name='订单明细', index=False)
+            if prepaid_df is not None:
+                prepaid_df.to_excel(writer, sheet_name='本月预收账款', index=False)
    except Exception as e:
        print(f"Excel生成失败：{str(e)}", file=sys.stderr)
        sys.exit(1)
--- a/skills/vala-order-amortization-stat/sql/detail.sql
+++ b/skills/vala-order-amortization-stat/sql/detail.sql
@ -195,24 +195,71 @@ SELECT
    WHEN DATE(amortization_start_date) > '{period_end}'::date THEN '试用期中'
    ELSE '正常订单'
  END AS "订单分类",
-  ROUND(pay_amount::numeric, 2) AS "订单金额",
+  ROUND(pay_amount::numeric, 2) AS "原订单金额",
+  ROUND(total_refund_amount::numeric, 2) AS "退款金额",
+  ROUND((pay_amount - total_refund_amount)::numeric, 2) AS "退后金额",
  ROUND(tax_rate::numeric, 4) AS "税率",
  ROUND(tax_amount::numeric, 2) AS "原订单税额",
-  ROUND(after_tax_amount::numeric, 2) AS "原订单税后金额",
+  ROUND(
+    CASE
+      WHEN has_refund = 1 AND is_full_refund = 0
+        THEN (pay_amount - total_refund_amount) / (1 + tax_rate)
+      WHEN has_refund = 1 AND is_full_refund = 1
+        THEN 0
+      ELSE after_tax_amount
+    END::numeric, 2
+  ) AS "当前税后金额",
+  ROUND(
+    CASE
+      WHEN has_refund = 1 AND is_full_refund = 0
+        THEN (pay_amount - total_refund_amount) / (1 + tax_rate) * tax_rate
+      WHEN has_refund = 1 AND is_full_refund = 1
+        THEN 0
+      ELSE tax_amount
+    END::numeric, 2
+  ) AS "当前税额",
  original_cycle_days AS "原始均摊周期",
  total_cycle_days AS "退后均摊周期",

-  -- 历史均摊金额（基于原始日均摊）
-  ROUND((historical_amorted_days * original_daily_amort_amount)::numeric, 2) AS "历史均摊金额",
-
-  -- 冲销历史均摊（场景B/E：历史下单+账期内退费，排除试用期内退费）
+  -- 历史累计均摊金额（仅历史订单有值，账期内订单=0；账期内退费按退费前状态算）
+  -- 账期前部分退费：按退后日均摊×退后历史天数
+  -- 正常订单/账期内退费：按原始日均摊×历史天数
  ROUND(
-    CASE WHEN is_current_period_refund = 1 AND is_ordered_in_period = 0 AND historical_amorted_days > 0
+    CASE
+      WHEN is_ordered_in_period = 1 THEN 0
+      WHEN is_pre_period_partial_refund = 1 AND historical_amorted_days > 0
+        THEN historical_amorted_days * daily_amort_amount
+      WHEN historical_amorted_days > 0
+        THEN historical_amorted_days * original_daily_amort_amount
+      ELSE 0
+    END::numeric, 2
+  ) AS "历史累计均摊金额",
+
+  -- 冲销历史累计摊销金额（仅场景B/E：历史下单+账期内退费，排除试用期内退费）
+  ROUND(
+    CASE WHEN is_current_period_refund = 1 AND is_ordered_in_period = 0
              AND is_trial_period_refund = 0
+              AND historical_amorted_days > 0
      THEN -(historical_amorted_days * original_daily_amort_amount)
      ELSE 0
    END::numeric, 2
-  ) AS "冲销历史均摊",
+  ) AS "冲销历史累计摊销金额",
+
+  -- 退后订单历史累计摊销金额（站在账期最后一天，看所有订单在账期第一天之前的有效历史摊销金额）
+  -- 账期内退费的订单按退后状态算；账期前部分退费按退后状态算；正常订单按原始状态算
+  ROUND(
+    CASE
+      WHEN is_ordered_in_period = 1 THEN 0
+      WHEN is_pre_period_partial_refund = 1 AND historical_amorted_days > 0
+        THEN historical_amorted_days * daily_amort_amount
+      WHEN is_current_period_refund = 1 AND is_ordered_in_period = 0
+           AND is_trial_period_refund = 0 AND historical_amorted_days > 0
+        THEN historical_amorted_days * daily_amort_amount
+      WHEN historical_amorted_days > 0
+        THEN historical_amorted_days * original_daily_amort_amount
+      ELSE 0
+    END::numeric, 2
+  ) AS "退后订单历史累计摊销金额",

  -- 冲销原税款（场景B/E：历史下单+账期内退费，排除试用期内退费）
  ROUND(
@ -232,7 +279,7 @@ SELECT
    END::numeric, 2
  ) AS "冲销未确认收入",

-  -- 账期均摊金额
+  -- 当期摊销金额
  ROUND(
    CASE
      -- 场景F：账期前全额退费，不统计
@ -274,7 +321,7 @@ SELECT
        END
      ELSE 0
    END::numeric, 2
-  ) AS "账期均摊金额",
+  ) AS "当期摊销金额",

  -- 退后订单税额（仅账期内下单且已转正的非全额退费订单计税）
  ROUND(
--- a/skills/vala-order-amortization-stat/sql/prepaid.sql
+++ b/skills/vala-order-amortization-stat/sql/prepaid.sql
@ -0,0 +1,148 @@
+-- 本月预收账款（Sheet3）
+-- 参数：{period_start} 账期起始日，{period_end} 账期结束日
+-- "本月" = 账期所在月，"上月" = 账期月前一个月
+-- 金额均为含税金额
+
+WITH
+-- 上月月份范围
+month_range AS (
+  SELECT
+    '{period_start}'::date AS period_start,
+    '{period_end}'::date AS period_end,
+    DATE_TRUNC('month', '{period_start}'::date) AS this_month_start,
+    (DATE_TRUNC('month', '{period_start}'::date) - INTERVAL '1 day')::date AS last_month_end,
+    (DATE_TRUNC('month', '{period_start}'::date) - INTERVAL '1 month')::date AS last_month_start
+),
+-- 所有退费记录
+all_refund_records AS (
+  SELECT
+    out_trade_no AS order_no,
+    SUM(refund_amount_int) / 100 AS total_refund_amount,
+    MAX(CASE WHEN refund_type = 2 THEN 1 ELSE 0 END) AS is_full_refund,
+    MAX(DATE(updated_at)) AS latest_refund_date
+  FROM bi_refund_order
+  WHERE status = 3
+  GROUP BY out_trade_no
+),
+-- 所有订单基础信息（含退费标记）
+order_base AS (
+  SELECT
+    o.out_trade_no AS order_no,
+    DATE(o.created_at) AS order_date,
+    o.pay_amount_int / 100 AS pay_amount,
+    CASE WHEN ar.order_no IS NOT NULL THEN 1 ELSE 0 END AS has_refund,
+    COALESCE(ar.total_refund_amount, 0) AS total_refund_amount,
+    CASE WHEN ar.order_no IS NOT NULL AND COALESCE(ar.total_refund_amount, 0) >= o.pay_amount_int / 100 THEN 1 ELSE 0 END AS is_full_refund,
+    ar.latest_refund_date,
+    -- 转正日期 = 下单日 + 7天
+    DATE(o.created_at) + INTERVAL '7 days' AS amortization_start_date,
+    -- 有效金额（部分退费 = 原金额 - 退费；未退费 = 原金额；全额退费 = 0）
+    CASE
+      WHEN ar.order_no IS NOT NULL AND COALESCE(ar.total_refund_amount, 0) >= o.pay_amount_int / 100 THEN 0
+      ELSE o.pay_amount_int / 100 - COALESCE(ar.total_refund_amount, 0)
+    END AS effective_amount,
+    o.order_status
+  FROM bi_vala_order o
+  JOIN bi_vala_app_account a ON o.account_id = a.id
+  LEFT JOIN all_refund_records ar ON o.out_trade_no = ar.order_no
+  CROSS JOIN month_range mr
+  WHERE
+    o.created_at >= mr.last_month_start
+    AND o.created_at <= mr.period_end
+    AND o.order_status IN (3,4)
+    AND o.pay_amount_int >= 1000
+    AND a.status = 1
+),
+-- 标记订单分类
+order_classified AS (
+  SELECT
+    ob.*,
+    mr.*,
+    -- 上月下单
+    CASE WHEN ob.order_date >= mr.last_month_start AND ob.order_date <= mr.last_month_end THEN 1 ELSE 0 END AS is_last_month_order,
+    -- 本月下单
+    CASE WHEN ob.order_date >= mr.this_month_start AND ob.order_date <= mr.period_end THEN 1 ELSE 0 END AS is_this_month_order,
+    -- 上月未转正（转正日 > 上月最后一天）
+    CASE WHEN ob.amortization_start_date > mr.last_month_end THEN 1 ELSE 0 END AS not_formal_in_last_month,
+    -- 本月已转正（转正日 <= 账期结束日）
+    CASE WHEN ob.amortization_start_date <= mr.period_end THEN 1 ELSE 0 END AS formal_in_this_month,
+    -- 上月已全额退费
+    CASE WHEN ob.has_refund = 1 AND ob.is_full_refund = 1 AND ob.latest_refund_date <= mr.last_month_end THEN 1 ELSE 0 END AS full_refund_in_last_month,
+    -- 本月退费金额（退费日期在账期内）
+    CASE WHEN ob.has_refund = 1
+              AND ob.latest_refund_date >= mr.period_start
+              AND ob.latest_refund_date <= mr.period_end
+         THEN ob.total_refund_amount
+         ELSE 0
+    END AS refund_in_period
+  FROM order_base ob
+  CROSS JOIN month_range mr
+)
+SELECT
+  '上月订单转正' AS "分类",
+  -- 新增：上月下单 + 上月未转正 + 剔除上月全额退费 + 部分退费算有效金额
+  ROUND(SUM(
+    CASE WHEN is_last_month_order = 1
+              AND not_formal_in_last_month = 1
+              AND full_refund_in_last_month = 0
+         THEN effective_amount
+         ELSE 0
+    END
+  )::numeric, 2) AS "新增",
+  -- 退费：这些订单在本月的退费金额
+  ROUND(SUM(
+    CASE WHEN is_last_month_order = 1
+              AND not_formal_in_last_month = 1
+              AND full_refund_in_last_month = 0
+         THEN refund_in_period
+         ELSE 0
+    END
+  )::numeric, 2) AS "退费"
+
+FROM order_classified
+
+UNION ALL
+
+SELECT
+  '本月订单转正' AS "分类",
+  -- 新增：本月下单 + 本月已转正 + 不考虑退费（原订单金额）
+  ROUND(SUM(
+    CASE WHEN is_this_month_order = 1
+              AND formal_in_this_month = 1
+         THEN pay_amount
+         ELSE 0
+    END
+  )::numeric, 2) AS "新增",
+  -- 退费：这些订单在本月的退费金额
+  ROUND(SUM(
+    CASE WHEN is_this_month_order = 1
+              AND formal_in_this_month = 1
+         THEN refund_in_period
+         ELSE 0
+    END
+  )::numeric, 2) AS "退费"
+
+FROM order_classified
+
+UNION ALL
+
+SELECT
+  '本月订单未转正' AS "分类",
+  -- 新增：本月下单 + 本月未转正 + 不考虑退费
+  ROUND(SUM(
+    CASE WHEN is_this_month_order = 1
+              AND formal_in_this_month = 0
+         THEN pay_amount
+         ELSE 0
+    END
+  )::numeric, 2) AS "新增",
+  -- 退费：这些订单在本月的退费金额
+  ROUND(SUM(
+    CASE WHEN is_this_month_order = 1
+              AND formal_in_this_month = 0
+         THEN refund_in_period
+         ELSE 0
+    END
+  )::numeric, 2) AS "退费"
+
+FROM order_classified;
--- a/skills/vala-order-amortization-stat/sql/summary.sql
+++ b/skills/vala-order-amortization-stat/sql/summary.sql
@ -122,66 +122,77 @@ order_with_days AS (
      THEN EXTRACT(DAY FROM (LEAST('{period_end}'::date, oc2.amortization_end_date) - oc2.amortization_start_date))::int + 1 ELSE 0
    END AS refunded_amort_days_to_period_end
  FROM order_classified oc2
-),
-- 步骤6：汇总
-summary_calc AS (
-  SELECT
-    -- 账期内新增订单统计
-    COUNT(CASE WHEN order_date >= '{period_start}'::date AND order_date <= '{period_end}'::date THEN 1 END) AS total_new_orders,
+)
+-- 步骤6：最终汇总（与订单明细Sheet2列对齐，前四列替换为订单统计+税率，其余SUM）
+SELECT
+  COUNT(CASE WHEN order_date >= '{period_start}'::date AND order_date <= '{period_end}'::date THEN 1 END) AS "订单数",
  COUNT(CASE WHEN order_date >= '{period_start}'::date AND order_date <= '{period_end}'::date
-                    AND amortization_start_date <= '{period_end}'::date THEN 1 END) AS formal_orders,
+                  AND amortization_start_date <= '{period_end}'::date THEN 1 END) AS "正式订单数",
  COUNT(CASE WHEN order_date >= '{period_start}'::date AND order_date <= '{period_end}'::date
-                    AND amortization_start_date > '{period_end}'::date THEN 1 END) AS trial_orders,
+                  AND amortization_start_date > '{period_end}'::date THEN 1 END) AS "试用订单数",

-    -- 冲销历史均摊（场景B/E：历史下单+账期内退费，排除试用期内退费）
+  -- 冲销历史累计摊销金额（SUM，明细列逻辑）
  ROUND(SUM(
-      CASE WHEN is_current_period_refund = 1 AND is_ordered_in_period = 0 AND historical_amorted_days > 0
+    CASE WHEN is_current_period_refund = 1 AND is_ordered_in_period = 0
              AND is_trial_period_refund = 0
+              AND historical_amorted_days > 0
+      THEN -(historical_amorted_days * original_daily_amort_amount)
+      ELSE 0
+    END
+  )::numeric, 2) AS "冲销历史累计摊销金额",
+
+  -- 退后订单历史累计摊销金额（SUM，明细列逻辑）
+  ROUND(SUM(
+    CASE
+      WHEN is_ordered_in_period = 1 THEN 0
+      WHEN is_pre_period_partial_refund = 1 AND historical_amorted_days > 0
+        THEN historical_amorted_days * daily_amort_amount
+      WHEN is_current_period_refund = 1 AND is_ordered_in_period = 0
+           AND is_trial_period_refund = 0 AND historical_amorted_days > 0
+        THEN historical_amorted_days * daily_amort_amount
+      WHEN historical_amorted_days > 0
        THEN historical_amorted_days * original_daily_amort_amount
      ELSE 0
    END
-    )::numeric, 2) AS writeoff_amort,
+  )::numeric, 2) AS "退后订单历史累计摊销金额",

-    -- 冲销原税款（场景B/E：历史下单+账期内退费，排除试用期内退费）
+  -- 冲销原税款（SUM）
  ROUND(SUM(
    CASE WHEN is_current_period_refund = 1 AND is_ordered_in_period = 0
              AND is_trial_period_refund = 0
-        THEN tax_amount
+      THEN -tax_amount
      ELSE 0
    END
-    )::numeric, 2) AS writeoff_tax,
+  )::numeric, 2) AS "冲销原税款",

-    -- 冲销未确认收入（场景B/E：历史下单+账期内退费，排除试用期内退费）
+  -- 冲销未确认收入（SUM）
  ROUND(SUM(
    CASE WHEN is_current_period_refund = 1 AND is_ordered_in_period = 0
              AND is_trial_period_refund = 0
-        THEN after_tax_amount - historical_amorted_days * original_daily_amort_amount
+      THEN -(after_tax_amount - historical_amorted_days * original_daily_amort_amount)
      ELSE 0
    END
-    )::numeric, 2) AS writeoff_unconfirmed,
+  )::numeric, 2) AS "冲销未确认收入",

-    -- 账期均摊金额
+  -- 当期摊销金额（SUM）
  ROUND(SUM(
    CASE
      WHEN is_pre_period_full_refund = 1 THEN 0
-        WHEN is_trial_period_refund = 1 AND is_full_refund = 1 THEN 0  -- 试用期内全额退费，无均摊
-        WHEN is_trial_period_refund = 1 AND is_full_refund = 0 THEN    -- 试用期内部分退费，按退后金额从转正日正常均摊
+      WHEN is_trial_period_refund = 1 AND is_full_refund = 1 THEN 0
+      WHEN is_trial_period_refund = 1 AND is_full_refund = 0 THEN
        CASE WHEN total_cycle_days > 0 AND period_amort_days > 0
          THEN daily_amort_amount * period_amort_days ELSE 0 END
-        WHEN is_current_period_refund = 1 AND is_full_refund = 1 THEN 0
-        -- 场景A：账期内下单+部分退费
+      WHEN is_current_period_refund = 1 AND is_full_refund = 1 AND is_ordered_in_period = 1 THEN 0
+      WHEN is_current_period_refund = 1 AND is_full_refund = 1 AND is_ordered_in_period = 0 THEN 0
      WHEN is_current_period_refund = 1 AND is_full_refund = 0 AND is_ordered_in_period = 1 THEN
        CASE WHEN total_cycle_days > 0 AND period_amort_days > 0
          THEN daily_amort_amount * period_amort_days ELSE 0 END
-        -- 场景B：历史下单+账期内部分退费，从转正日重新均摊
      WHEN is_current_period_refund = 1 AND is_full_refund = 0 AND is_ordered_in_period = 0 THEN
        CASE WHEN total_cycle_days > 0 AND refunded_amort_days_to_period_end > 0
          THEN daily_amort_amount * refunded_amort_days_to_period_end ELSE 0 END
-        -- 场景C：账期前部分退费
      WHEN is_pre_period_partial_refund = 1 THEN
        CASE WHEN total_cycle_days > 0 AND period_amort_days > 0
          THEN daily_amort_amount * period_amort_days ELSE 0 END
-        -- 正常订单
      WHEN period_amort_days > 0 THEN
        CASE
          WHEN eff_end = amortization_end_date
@ -190,37 +201,36 @@ summary_calc AS (
        END
      ELSE 0
    END
-    )::numeric, 2) AS period_amort,
+  )::numeric, 2) AS "当期摊销金额",

-    -- 账期税费（仅账期内下单且已转正的非全额退费订单）
+  -- 账期税额（SUM）
  ROUND(SUM(
    CASE
-        -- 账期内下单 + 已转正 + 部分退费 → 退后订单税额
      WHEN is_ordered_in_period = 1
-             AND amortization_start_date <= '{period_end}'::date
+           AND DATE(amortization_start_date) <= '{period_end}'::date
           AND is_current_period_refund = 1 AND is_full_refund = 0
      THEN refunded_tax_amount
-        -- 账期内下单 + 已转正 + 未全额退费（含未退费） → 原订单税额
      WHEN is_ordered_in_period = 1
-             AND amortization_start_date <= '{period_end}'::date
+           AND DATE(amortization_start_date) <= '{period_end}'::date
           AND (has_refund = 0 OR (is_current_period_refund = 0 AND is_full_refund = 0))
      THEN tax_amount
      ELSE 0
    END
-    )::numeric, 2) AS period_tax
+  )::numeric, 2) AS "账期税额",

-  FROM order_with_days
-  WHERE is_pre_period_full_refund = 0
-)
-SELECT
-  total_new_orders AS "订单数",
-  formal_orders AS "正式订单数",
-  trial_orders AS "试用订单数",
-  writeoff_amort AS "冲销历史均摊",
-  writeoff_tax AS "冲销原税款",
-  writeoff_unconfirmed AS "冲销未确认收入",
-  period_amort AS "账期均摊金额",
-  period_tax AS "账期税费",
-  ROUND((period_amort - writeoff_amort)::numeric, 2) AS "冲销后均摊金额",
-  ROUND((period_tax - writeoff_tax)::numeric, 2) AS "冲销后税费"
-FROM summary_calc;
+  -- 未确认收入（SUM）
+  ROUND(SUM(
+    CASE
+      WHEN is_pre_period_full_refund = 1 THEN 0
+      WHEN is_current_period_refund = 1 AND is_full_refund = 1 THEN 0
+      WHEN is_current_period_refund = 1 AND is_full_refund = 0 THEN
+        refunded_after_tax_amount - daily_amort_amount * refunded_amort_days_to_period_end
+      WHEN is_pre_period_partial_refund = 1 THEN
+        refunded_after_tax_amount - daily_amort_amount * total_amorted_days_to_period_end
+      ELSE
+        after_tax_amount - original_daily_amort_amount * total_amorted_days_to_period_end
+    END
+  )::numeric, 2) AS "未确认收入"
+
+FROM order_with_days
+WHERE is_pre_period_full_refund = 0;