diff --git a/.gitignore b/.gitignore
index 6105287..da062e8 100644
--- a/.gitignore
+++ b/.gitignore
@@ -6,3 +6,59 @@ __pycache__/
 *.jpg
 *.png
 .localized
+
+# Raw local downloads: retain only curated production evidence below.
+*.xlsx
+*.xls
+*.zip
+*.7z
+*.rar
+*.tar
+*.gz
+*.dmg
+*.pkg
+*.exe
+*.pdf
+*.doc
+*.docx
+*.rtf
+
+# Production monthly-assessment evidence retained in this repository.
+!20260713/**/*.xlsx
+!20260713_1500/**/*.xlsx
+!5月18T测算/**/*.xlsx
+!6月18T测算/**/*.xlsx
+!6月核算结果/**/*.xlsx
+!7月核算结果/**/*.xlsx
+!7月绩效重算_多源考核里程_20260817/**/*.xlsx
+!租赁任务考核_*.xlsx
+!里程任务考核_*.xlsx
+!*干预情况说明*.xlsx
+!docs/**
+!**/*考核*.pdf
+!**/*考核*.doc
+!**/*考核*.docx
+!**/*核算*.pdf
+!**/*核算*.doc
+!**/*核算*.docx
+!**/*测试*.pdf
+!**/*测试*.doc
+!**/*测试*.docx
+!**/*验证*.pdf
+!**/*验证*.doc
+!**/*验证*.docx
+!**/*说明*.pdf
+!**/*说明*.doc
+!**/*说明*.docx
+
+# Never commit credentials or local secret material.
+.env
+.env.*
+*accesskey*
+*AccessKey*
+*secret*
+*token*
+*credential*
+*password*
+*private*key*
+*backup*code*
diff --git a/01-可行性与实施计划书.md b/01-可行性与实施计划书.md
new file mode 100644
index 0000000..56c20ed
--- /dev/null
+++ b/01-可行性与实施计划书.md
@@ -0,0 +1,178 @@
+# 养馋记物流合作 · 可行性与实施计划书
+
+| 项 | 内容 |
+|----|------|
+| 版本 | V0.1 |
+| 日期 | 2026-07-31 |
+| 状态 | 评估稿（待管理层战略拍板；单价/税点待书面确认） |
+| 配套 | [流程图集](./02-流程图集.md) · [分期实施计划](./03-分期实施计划.md) |
+| 依据 | 养馋记合作交流纪要；`self-operated-logistics/assumptions.md`；`物流自营-MVP与版本计划.md`；`物流自营盈亏-粗方案-会议纪要落地.md`；`物流项目立项-合同交车衔接方案.md` |
+
+---
+
+## 1. 背景与问题
+
+养馋记加快外区扩张，倾向**只选一家主物流商**，价格敏感，对标对象为顺丰（网络、回程、昼夜复用、单车多趟）。  
+羚牛已参与项目运营并具备氢能协同价值，但报价与综合成本仍有差距；差距主因不是「有没有车」，而是：
+
+1. 外区回程空驶  
+2. 车辆全日利用率不足  
+3. 线路组合与系统化排线能力不足  
+4. 战略定位（短期利润 vs 规模客户）未内部拍板  
+
+同时会议提出 TMS / 智能调度诉求。现有 OneOS 物流条线已锁定为**轻量调度、非 TMS**，中期靶子是「合同 → 交车 → 出账 → 证据 → 盈亏」。
+
+---
+
+## 2. 计划目标
+
+| 目标 | 说明 |
+|------|------|
+| G1 决策可下 | 管理层明确 Go / No-Go / 有条件 Go 及利润底线 |
+| G2 主链可挂 | 养馋记作为物流项目接入现有主链，能出可解释账单 |
+| G3 成本可谈 | 全成本与利用率、空驶可视，支撑对客报价 |
+| G4 运力可扩 | 回程与昼夜复用有试点与台阶承诺，匹配门店扩张 |
+| G5 系统不越界 | 不对外承诺「上线即对标顺丰 TMS」；算法另立项 |
+
+**非目标：** 用绿色品牌替代价格竞争力；另起独立「养馋记 TMS」产品线抢占物流 MVP 资源。
+
+---
+
+## 3. 可行性结论
+
+### 3.1 总判
+
+| 维度 | 结论 |
+|------|------|
+| 商务/运营 | **有条件可行**——前提是回程走廊可拼、战略底线清晰、运力台阶可承诺 |
+| 产品近中期 | **可行且应做**——经营底座 + 利用率/空驶分析，挂现有物流 MVP/V0.2 |
+| 智能 TMS/算法 | **中长期可行、短期不可当作商务武器**——需专班与真实货源数据 |
+
+### 3.2 衔接总策略（一句话）
+
+**客户项目化接入现有物流主链；运营先解决货源与复用；系统先记账再调度台，最后才算法。**
+
+```text
+计价规则 → 立项 → 合同生效 → 交车 → 轻量调度 → 趟/段出账+证据 → 成本T+1 → 项目盈亏
+                ↑
+         养馋记：规则包 + 门店线路主数据 + 多趟/昼夜/空驶/返程字段
+                ↑
+         跨项目：人工排班 → 占用日历 →（后期）回程候选推荐
+```
+
+### 3.3 与顺丰对比（决策用）
+
+| 能力 | 顺丰 | 羚牛现状/路径 |
+|------|------|----------------|
+| 区域网络与回程货 | 强 | 弱 → 靠现有项目走廊匹配 |
+| 昼夜业务组合 | 成熟 | 需运营拼线 + 占用日历 |
+| 单车多趟 | 常态 | 字段支持 + 调度台后置 |
+| 单公里报价 | 未必最低 | 忌纯砍价 |
+| 氢能/绿色 | 非其主叙事 | 羚牛差异化加分，不替代运价 |
+| 系统排线 | 成熟网络调度 | 我方 S3/S4，不绑 S0 商务承诺 |
+
+---
+
+## 4. 范围与边界
+
+### 4.1 范围内
+
+- 养馋记项目：规则、合同、交车、调度任务、出账、证据、盈亏  
+- 主数据：门店、线路、频次、时间窗、货量、车型  
+- 经营指标：去程/返程/空驶、单车趟次与利用率、线路成本  
+- 跨项目占用（先人工后系统提示）  
+- 氢能第二账（能源/财务，与物流盈亏分表）  
+- 分期评估调度台与智能优化（另预算）
+
+### 4.2 范围外（本期计划书明确不做）
+
+- 完整 VRP/动态重调度作为商务交付承诺  
+- 承运商招标平台、客户系统 API 强对接（可单独立项）  
+- 租赁提车应收 / 还车应结混入物流资金流  
+- 推翻「非 TMS」假设、用演示算法换独家合同  
+
+### 4.3 与现版本对齐
+
+| 物流版本 | 本计划对应 |
+|----------|------------|
+| MVP V0.1 | 阶段 S1 |
+| V0.2 | 阶段 S2 |
+| V0.3 规则/立项 | 规则包正式版、立项挂接 |
+| V0.4 GPS | 在途可视（非智能排线） |
+| 新立项 | 阶段 S3 调度台、S4 算法 |
+
+---
+
+## 5. 成本与投入（量级）
+
+> 非正式报价；含产品研发量级，不含车辆购置、司机薪酬、商务让利。
+
+| 阶段 | 内容 | 量级 | 费用感 |
+|------|------|------|--------|
+| S0 | 全成本、顺丰差拆解、回程走廊、战略底线 | 业务财务为主 · 0.5～1 人周～人月 | 很低 |
+| S1 | MVP + 养馋规则/模板/空驶多趟字段 | 增量 1～2 人月 | 低 |
+| S2 | 验真、成本挂接、利用率报表、占用日历 | 2～4 人月 | 中低 |
+| S3 | 半自动调度台 | 4～8 人月 | 中 |
+| S4 | 智能排线/回程推荐 | 6～12+ 人月 + 算法专班跨季 | 高 |
+
+**隐性成本：** 回程货开发人力、薄利下备车、若过度承诺系统能力导致的信誉损失。
+
+**预算建议：** 先批 S0+S1；战略 Go 后批 S2；S3/S4 单独立项，不写入对客必达条款。
+
+---
+
+## 6. 风险与难点
+
+| 等级 | 风险/难点 | 应对 |
+|------|-----------|------|
+| 高 | 回程货源拼不成仍接独家 | S0 门禁；溢价或收缩范围 |
+| 高 | 战略未拍板却锁低价长约 | 先内部底线再报价 |
+| 中 | 系统预期=顺丰 TMS | 对外话术与分期绑定；禁演示忽悠 |
+| 中 | 主数据（门店窗货量）缺失 | S3 入口条件 |
+| 中 | 运力跟不上外区开店 | 台阶式运力承诺 |
+| 低 | 多客户导入模板差异 | 规则中心+映射（既有规划） |
+| 低 | 氢能归因不清 | 物流账 / 能源账分表 |
+
+---
+
+## 7. 组织与职责（RACI 摘要）
+
+| 事项 | 业务 | 财务 | 物流运营 | 能源 | 数智/产品 | 管理层 |
+|------|------|------|----------|------|-----------|--------|
+| 战略定位与底线 | C | C | I | C | I | **A/R** |
+| 全成本与报价 | R | **A** | C | C | I | C |
+| 回程走廊匹配 | C | I | **A/R** | I | C | I |
+| 主数据收集 | R | I | C | I | C | I |
+| S1–S2 系统 | C | C | C | C | **A/R** | I |
+| S3–S4 立项 | C | C | C | I | **R** | **A** |
+| 对客方案 | **R** | C | C | C | C | A |
+
+A=批准 R=执行 C=协商 I=知会
+
+---
+
+## 8. 成功标准
+
+1. 管理层形成书面 Go / No-Go / 有条件 Go。  
+2. 任意试点趟次可回答：合同、车、司机、门店序列、收入分项、空驶公里、成本粗算。  
+3. 回程走廊表覆盖主要外区线路，可拼比例与不可拼溢价策略清晰。  
+4. 对内汇报不出现「已具备顺丰级智能排线」类越界表述。  
+5. S2 起单车日趟次、利用率、空驶率可周更。
+
+---
+
+## 9. 决策建议
+
+1. **推荐路径：** 有条件推进——S0 过门 → 挂主链（S1）→ 经营可解释（S2）→ 视回程试点再开 S3/S4。  
+2. **不推荐：** 以智能 TMS 承诺换独家；或纯降单公里硬刚顺丰。  
+3. **差异化：** 运营熟悉度 + 氢能规模第二账 + 极简透明合作关系（对齐养馋记供应商理念）。  
+
+待拍板清单见 [实施计划 §6](./03-分期实施计划.md#6-待拍板清单)。
+
+---
+
+## 10. 修订记录
+
+| 版本 | 日期 | 说明 |
+|------|------|------|
+| V0.1 | 2026-07-31 | 首版：计划书 + 流程图 + 实施计划成套 |
diff --git a/0813.md b/0813.md
new file mode 100644
index 0000000..c3c9fdf
--- /dev/null
+++ b/0813.md
@@ -0,0 +1,81 @@
+# 养馋记物流派单与对账模块建设评估
+
+> **管理层决策版｜本次仅申请第一阶段立项，不进入全量开发承诺**
+
+## 一、结论与建议
+
+### 结论
+
+**项目可做，建议分三阶段推进。** 先以养馋记订单与路线数据接入 OneOS，跑通“订单—人工派单—运输办结—费用归集—T+1 暂算盈亏”闭环；月度对账和 AI 智能调度分别后置，避免在业务规则、接口和数据尚未验证前一次性投入。
+
+### 本次建议确认
+
+1. **批准第一阶段 MVP**：周期 1 个月；人力成本 **11.4 万元**，首年全口径预算 **15.78 万元**
+2. **批准启动接口与业务确认**：API/字段、费用试算样例、OneOS 成本匹配规则确认后正式开工
+3. **第二、三阶段按门禁决策**：第一阶段跑通并验证数据口径后，再确认第二阶段投入；AI 不纳入本次预算
+
+## 二、第一阶段：一个月先解决什么
+
+| 管理层关心的问题 | 第一阶段交付 |
+| --- | --- |
+| 订单能否自动进入系统 | 对接养馋记 API；接口异常时支持 Excel/人工建单兜底 |
+| 调度能否更快、更可控 | 自动生成待派任务，调度人工选择车辆和司机；司机端接收、执行和办结任务 |
+| 每笔单是否赚钱 | 归集运费、OneOS 氢费和 ETC，形成逐单暂算；支持 **T+1 查看前一日收入、成本和盈亏** |
+| 本期不做什么 | 完整月度对账、合作方在线确认/驳回、用友接口、全自动派单及 AI 调度 |
+
+**第一阶段验收结果：** 养馋记订单可进入 OneOS，调度可完成派车派司机，司机可办结任务，管理层可在次日看到单据层面的暂算盈亏。
+
+## 三、分阶段投入与边界
+
+| 阶段 | 目标与主要交付 | 周期 | 阶段人力成本 | 决策边界 |
+| --- | --- | ---: | ---: | --- |
+| **第一阶段：MVP** | 订单接入、人工派单、司机执行、基础费用归集、T+1 暂算盈亏 | 1 个月 | **11.4 万元** | **本次申请批准** |
+| **第二阶段：业务与对账闭环** | 改单取消与异常处理、完整费用及车辆成本、利润精算、月度账单、合作方在线对账 | 1—2 个月 | **11.4—22.8 万元** | 第一阶段验证通过后启动；累计 **22.8—34.2 万元** |
+| **第三阶段：AI 智能调度试点** | 基于真实调度数据，试点串单/串点、复跑、车型匹配和成本优化建议 | 数据积累 2—3 个月；试点开发 1—2 个月 | **11.4—28.8 万元** | 单独立项；含 AI 的远期累计 **34.2—63.0 万元** |
+
+## 四、成本口径与首年全口径预算
+
+### 人力成本口径
+
+- 第一、二阶段：产品 1 人、研发 2 人、测试 1 人
+- 产品、研发 **1,500 元/人天**；测试 **1,200 元/人天**
+- 按 20 个工作日/月测算，满配日成本 **5,700 元**
+- AI 试点按产品 1 人、研发/算法 2—3 人、测试 1 人测算；数据积累随前两阶段实际运营完成，不单列开发人力
+
+### 非人力成本预算（首年）
+
+| 成本项 | 预算口径 | 首年预算 |
+| --- | --- | ---: |
+| 应用服务器 | 4核8G、100G 云盘、5Mbps 公网带宽，**2 台**（生产/测试或应用隔离） | **12,000 元** |
+| 托管数据库 | RDS MySQL 2核8G、50G 存储、基础备份 | **6,000 元** |
+| 对象存储、备份与日志 | 签收凭证、导入文件、备份、日志及基础流量 | **2,000 元** |
+| 地图、路径与 GPS 服务 | 低量调用预留；优先复用 OneOS 现有车辆 GPS 数据 | **6,000 元** |
+| 短信/消息 | 按 10,000 条通知短信预留 | **500 元** |
+| 小程序认证 | 微信小程序主体认证 | **300 元** |
+| 域名、SSL、安全监控 | 域名、证书、基础监控与告警 | **2,000 元** |
+| 养馋记 API 联调/信息服务 | 对方可能收取的接口支持及联调服务费预留，待对方确认 | **10,000 元** |
+| 上线部署与基础运维配置 | 环境初始化、备份策略、发布与巡检配置 | **5,000 元** |
+| **非人力成本合计** | 其中年度运行成本约 28,800 元；一次性启动预留 15,000 元 | **43,800 元** |
+
+### 项目首年投入总览
+
+| 决策范围 | 人力成本 | 非人力成本 | 首年总投入 |
+| --- | ---: | ---: | ---: |
+| **本次：第一阶段 MVP** | 114,000 元 | 43,800 元 | **157,800 元** |
+| 第一、二阶段完成 | 228,000—342,000 元 | 43,800 元 | **271,800—385,800 元** |
+| 含 AI 试点的远期累计 | 342,000—630,000 元 | 43,800 元 | **385,800—673,800 元** |
+
+**说明：** 上述为预算，不是供应商最终报价；云资源价格会受地域、带宽、磁盘和活动折扣影响。地图/GPS 预算基于低量调用并复用现有 OneOS 数据；如需单独采购地图商用授权、高频路径规划或新增 GPS 设备，应单独询价。用友接口及外部 AI 模型调用费用本期均不包含。
+
+## 五、启动门禁与风险控制
+
+| 启动前必须确认 | 不确认的影响 | 控制方式 |
+| --- | --- | --- |
+| API 文档、沙箱、可提供字段及同步时效 | 订单接入范围和工期无法锁定 | 先完成接口清单和联调验证；Excel/人工建单作为兜底 |
+| 新单、改单、取消单及状态回传规则 | 运单状态和对账口径可能不一致 | 以业务样例确认状态规则后开发 |
+| 运费、抽点、杂费、车辆成本试算样例 | 暂算与最终账可能存在差异 | 第一阶段先做暂算，第二阶段固化完整计费与对账规则 |
+| OneOS 氢费、ETC 数据来源与匹配规则 | 无法形成可信的成本与利润 | 明确数据来源、匹配频率和异常处理规则 |
+
+## 六、建议
+
+**建议本次批准第一阶段首年全口径预算 15.78 万元**（人力 11.4 万元 + 非人力 4.38 万元）。用一个月验证接口可用性、人工调度闭环和 T+1 盈亏价值；通过后再投入第二阶段的对账闭环。AI 智能调度必须在真实订单、车辆、司机与成本数据积累 2—3 个月后单独评估，避免“先投入、后找场景”。
diff --git a/20260713/离职人员6月奖金处理明细.xlsx b/20260713/离职人员6月奖金处理明细.xlsx
new file mode 100644
index 0000000..3721314
Binary files /dev/null and b/20260713/离职人员6月奖金处理明细.xlsx differ
diff --git a/20260713/租赁任务考核_2026年6月.xlsx b/20260713/租赁任务考核_2026年6月.xlsx
new file mode 100644
index 0000000..8470b1f
Binary files /dev/null and b/20260713/租赁任务考核_2026年6月.xlsx differ
diff --git a/20260713/租赁任务考核_2026年6月_干预情况说明.xlsx b/20260713/租赁任务考核_2026年6月_干预情况说明.xlsx
new file mode 100644
index 0000000..d652542
Binary files /dev/null and b/20260713/租赁任务考核_2026年6月_干预情况说明.xlsx differ
diff --git a/20260713/里程任务考核_6月核算_2.xlsx b/20260713/里程任务考核_6月核算_2.xlsx
new file mode 100644
index 0000000..e4ca7d5
Binary files /dev/null and b/20260713/里程任务考核_6月核算_2.xlsx differ
diff --git a/里程考核绩效核算/数据来源/1月.xlsx b/20260713_1500/1月.xlsx
similarity index 100%
rename from 里程考核绩效核算/数据来源/1月.xlsx
rename to 20260713_1500/1月.xlsx
diff --git a/里程考核绩效核算/数据来源/2月.xlsx b/20260713_1500/2月.xlsx
similarity index 100%
rename from 里程考核绩效核算/数据来源/2月.xlsx
rename to 20260713_1500/2月.xlsx
diff --git a/里程考核绩效核算/数据来源/3月.xlsx b/20260713_1500/3月.xlsx
similarity index 100%
rename from 里程考核绩效核算/数据来源/3月.xlsx
rename to 20260713_1500/3月.xlsx
diff --git a/20260713_1500/6月考核数据去重说明.md b/20260713_1500/6月考核数据去重说明.md
new file mode 100644
index 0000000..6e4c620
--- /dev/null
+++ b/20260713_1500/6月考核数据去重说明.md
@@ -0,0 +1,76 @@
+# 2026年6月里程考核 — 源数据去重说明
+
+## 背景
+
+6月考核源数据从新系统同步时，由于部分车辆交还车流程未闭环，同一辆车同时存在两条考核任务记录。其中一条是早已还车产生的冗余记录。
+
+经排查确认 **6辆车存在重复导入**，旧源文件 429 行 → 去重后新源文件 423 行，共删除 6 条冗余记录。
+
+---
+
+## 涉及的6辆车
+
+| 车牌 | 销售 | 旧源 | 新源 | 考核目标 | 6月实际里程 | 是否达标 | 去重影响 |
+|------|------|------|------|----------|------------|----------|----------|
+| 粤AGP4335 | 刘念念 | 2条重复 | 1条 | 羚牛136辆4.5T冷链车 | 7,087 | 达标 | **奖金减半** |
+| 粤AGP5159 | 金可鹏 | 1条 | 0条(删除) | 交投190辆4.5T冷链车 | 7,769 | 达标 | **奖金消失** |
+| 粤AGP5685 | 金可鹏 | 1条 | 0条(删除) | 交投190辆4.5T冷链车 | 7,068 | 达标 | **奖金消失** |
+| 粤AGR8799 | 钟祥 | 1条 | 0条(删除) | 交投190辆4.5T冷链车 | 2,869 | 未达标 | 无影响(0奖金) |
+| 粤AGQ5808 | 金可鹏 | 2条重复 | 1条 | 交投190辆4.5T冷链车 | 1,360 | 未达标 | 无影响(0奖金) |
+| 粤AGR9896 | 金可鹏 | 2条重复 | 1条 | 交投190辆4.5T冷链车 | 1,242 | 未达标 | 无影响(0奖金) |
+
+> 后3辆虽重复，但考核结果为「未达标」，奖金为0，去重前后无差异。
+
+---
+
+## 发放金额差异（旧源含重复 vs 新源去重）
+
+### 总体
+
+| | 考核应发合计 |
+|------|------------|
+| 旧源（含6条重复） | **97,782.10** |
+| 新源（去重后） | **96,665.27** |
+| **多发金额** | **-1,116.83** |
+
+### 分摊到人
+
+| 销售 | 旧源(含重复) | 新源(去重) | 差异 | 说明 |
+|------|-------------|-----------|------|------|
+| **刘念念** | 32,590.31 | 32,027.25 | **-563.06** | 粤AGP4335 去重，当月520→260，补发1/2/4月消失 |
+| **金可鹏** | 17,985.33 | 17,331.56 | **-653.77** | 粤AGP5159+粤AGP5685 删除，当月+补发消失；连锁粤A05776F新增+33.33 |
+| **钟祥** | 2,928.02 | 3,028.02 | **+100.00** | 连锁反应：粤A01396F 累计变化后达标，新增当月100 |
+| 其余9人 | — | — | **0.00** | 完全一致 |
+
+### 逐条明细
+
+**旧源多发（去重后消失，共990.16元）：**
+
+| 车牌 | 销售 | 发放类型 | 多发金额 |
+|------|------|----------|----------|
+| 粤AGP4335 | 刘念念 | 当月（重复导致翻倍） | 260.00 |
+| 粤AGP4335 | 刘念念 | 补发1月 | 8.39 |
+| 粤AGP4335 | 刘念念 | 补发2月 | 260.00 |
+| 粤AGP4335 | 刘念念 | 补发4月 | 34.67 |
+| 粤AGP5159 | 金可鹏 | 当月 | 150.00 |
+| 粤AGP5685 | 金可鹏 | 当月 | 150.00 |
+| 粤AGP5685 | 金可鹏 | 补发1月 | 150.00 |
+| 粤AGP5685 | 金可鹏 | 补发2月 | 150.00 |
+| 粤AGP5685 | 金可鹏 | 补发3月 | 48.39 |
+| 粤AGP5685 | 金可鹏 | 补发5月 | 38.71 |
+
+**新源新增（连锁反应，共133.33元）：**
+
+| 车牌 | 销售 | 发放类型 | 新增金额 |
+|------|------|----------|----------|
+| 粤A01396F | 钟祥 | 当月 | 100.00 |
+| 粤A05776F | 金可鹏 | 当月 | 33.33 |
+
+---
+
+## 结论
+
+- 旧源因重复记录**多发 1,116.83 元**，涉及刘念念(-563.06)、金可鹏(-653.77)、钟祥(+100.00)
+- 新源去重后数据正确，以新源计算结果为准：**考核应发 96,665.27 元**
+- GPS数据源无问题（本地G7+东方北斗与数据库一致）
+- 其余9人发放金额完全不受影响
diff --git a/20260713_1500/6月考核数据去重说明.xlsx b/20260713_1500/6月考核数据去重说明.xlsx
new file mode 100644
index 0000000..bd521d4
Binary files /dev/null and b/20260713_1500/6月考核数据去重说明.xlsx differ
diff --git a/20260713_1500/离职人员6月奖金处理明细.xlsx b/20260713_1500/离职人员6月奖金处理明细.xlsx
new file mode 100644
index 0000000..e671ec5
Binary files /dev/null and b/20260713_1500/离职人员6月奖金处理明细.xlsx differ
diff --git a/里程考核绩效核算/数据来源/租赁任务考核_2026年1月.xlsx b/20260713_1500/租赁任务考核_2026年1月.xlsx
similarity index 100%
rename from 里程考核绩效核算/数据来源/租赁任务考核_2026年1月.xlsx
rename to 20260713_1500/租赁任务考核_2026年1月.xlsx
diff --git a/里程考核绩效核算/数据来源/租赁任务考核_2026年2月.xlsx b/20260713_1500/租赁任务考核_2026年2月.xlsx
similarity index 100%
rename from 里程考核绩效核算/数据来源/租赁任务考核_2026年2月.xlsx
rename to 20260713_1500/租赁任务考核_2026年2月.xlsx
diff --git a/里程考核绩效核算/数据来源/租赁任务考核_2026年3月.xlsx b/20260713_1500/租赁任务考核_2026年3月.xlsx
similarity index 100%
rename from 里程考核绩效核算/数据来源/租赁任务考核_2026年3月.xlsx
rename to 20260713_1500/租赁任务考核_2026年3月.xlsx
diff --git a/20260713_1500/租赁任务考核_2026年4月.xlsx b/20260713_1500/租赁任务考核_2026年4月.xlsx
new file mode 100644
index 0000000..64e0308
Binary files /dev/null and b/20260713_1500/租赁任务考核_2026年4月.xlsx differ
diff --git a/20260713_1500/租赁任务考核_2026年5月.xlsx b/20260713_1500/租赁任务考核_2026年5月.xlsx
new file mode 100644
index 0000000..d38f9fc
Binary files /dev/null and b/20260713_1500/租赁任务考核_2026年5月.xlsx differ
diff --git a/20260713_1500/租赁任务考核_2026年6月 (1).xlsx b/20260713_1500/租赁任务考核_2026年6月 (1).xlsx
new file mode 100644
index 0000000..11cfe00
Binary files /dev/null and b/20260713_1500/租赁任务考核_2026年6月 (1).xlsx differ
diff --git a/20260713_1500/租赁任务考核_2026年6月.xlsx b/20260713_1500/租赁任务考核_2026年6月.xlsx
new file mode 100644
index 0000000..aa6d92c
Binary files /dev/null and b/20260713_1500/租赁任务考核_2026年6月.xlsx differ
diff --git a/20260713_1500/租赁任务考核_2026年6月_干预情况说明.xlsx b/20260713_1500/租赁任务考核_2026年6月_干预情况说明.xlsx
new file mode 100644
index 0000000..e6b722f
Binary files /dev/null and b/20260713_1500/租赁任务考核_2026年6月_干预情况说明.xlsx differ
diff --git a/20260713_1500/里程任务考核_6月核算_2.xlsx b/20260713_1500/里程任务考核_6月核算_2.xlsx
new file mode 100644
index 0000000..92b5428
Binary files /dev/null and b/20260713_1500/里程任务考核_6月核算_2.xlsx differ
diff --git a/20260713_1500/里程任务考核_6月核算_（发放记录已删除离职人员补发和转接）.xlsx b/20260713_1500/里程任务考核_6月核算_（发放记录已删除离职人员补发和转接）.xlsx
new file mode 100644
index 0000000..a5c02af
Binary files /dev/null and b/20260713_1500/里程任务考核_6月核算_（发放记录已删除离职人员补发和转接）.xlsx differ
diff --git a/5月18T测算/方案1_原逻辑_1000基准.xlsx b/5月18T测算/方案1_原逻辑_1000基准.xlsx
new file mode 100644
index 0000000..f2d9c9a
Binary files /dev/null and b/5月18T测算/方案1_原逻辑_1000基准.xlsx differ
diff --git a/5月18T测算/方案1_原逻辑_1000基准_0701.xlsx b/5月18T测算/方案1_原逻辑_1000基准_0701.xlsx
new file mode 100644
index 0000000..9fe0fd6
Binary files /dev/null and b/5月18T测算/方案1_原逻辑_1000基准_0701.xlsx differ
diff --git a/5月18T测算/方案2_简化版_800_300仅当月.xlsx b/5月18T测算/方案2_简化版_800_300仅当月.xlsx
new file mode 100644
index 0000000..f760217
Binary files /dev/null and b/5月18T测算/方案2_简化版_800_300仅当月.xlsx differ
diff --git a/5月18T测算/方案2_简化版_800_300仅当月_0701.xlsx b/5月18T测算/方案2_简化版_800_300仅当月_0701.xlsx
new file mode 100644
index 0000000..1f061c6
Binary files /dev/null and b/5月18T测算/方案2_简化版_800_300仅当月_0701.xlsx differ
diff --git a/5月18T测算/方案3_完整版_800_300_500.xlsx b/5月18T测算/方案3_完整版_800_300_500.xlsx
new file mode 100644
index 0000000..20322fc
Binary files /dev/null and b/5月18T测算/方案3_完整版_800_300_500.xlsx differ
diff --git a/5月18T测算/方案3_完整版_800_300_500_0701.xlsx b/5月18T测算/方案3_完整版_800_300_500_0701.xlsx
new file mode 100644
index 0000000..967f2bc
Binary files /dev/null and b/5月18T测算/方案3_完整版_800_300_500_0701.xlsx differ
diff --git a/6月18T测算/6月18T_方案3.xlsx b/6月18T测算/6月18T_方案3.xlsx
new file mode 100644
index 0000000..d5fb3e7
Binary files /dev/null and b/6月18T测算/6月18T_方案3.xlsx differ
diff --git a/6月核算结果/离职人员6月奖金处理明细.xlsx b/6月核算结果/离职人员6月奖金处理明细.xlsx
new file mode 100644
index 0000000..6c454ab
Binary files /dev/null and b/6月核算结果/离职人员6月奖金处理明细.xlsx differ
diff --git a/6月核算结果/租赁任务考核_2026年6月.xlsx b/6月核算结果/租赁任务考核_2026年6月.xlsx
new file mode 100644
index 0000000..8470b1f
Binary files /dev/null and b/6月核算结果/租赁任务考核_2026年6月.xlsx differ
diff --git a/6月核算结果/租赁任务考核_2026年6月_干预情况说明.xlsx b/6月核算结果/租赁任务考核_2026年6月_干预情况说明.xlsx
new file mode 100644
index 0000000..d652542
Binary files /dev/null and b/6月核算结果/租赁任务考核_2026年6月_干预情况说明.xlsx differ
diff --git a/6月核算结果/里程任务考核_6月核算.xlsx b/6月核算结果/里程任务考核_6月核算.xlsx
new file mode 100644
index 0000000..eeae3cf
Binary files /dev/null and b/6月核算结果/里程任务考核_6月核算.xlsx differ
diff --git a/7月核算结果/7月考核车辆多源里程汇总.xlsx b/7月核算结果/7月考核车辆多源里程汇总.xlsx
new file mode 100644
index 0000000..93f7715
Binary files /dev/null and b/7月核算结果/7月考核车辆多源里程汇总.xlsx differ
diff --git a/7月核算结果/7月考核车辆多源里程汇总_GPS全量回补前.xlsx b/7月核算结果/7月考核车辆多源里程汇总_GPS全量回补前.xlsx
new file mode 100644
index 0000000..778bb23
Binary files /dev/null and b/7月核算结果/7月考核车辆多源里程汇总_GPS全量回补前.xlsx differ
diff --git a/7月核算结果/7月考核车辆多源里程汇总_含车型.xlsx b/7月核算结果/7月考核车辆多源里程汇总_含车型.xlsx
new file mode 100644
index 0000000..cb84b84
Binary files /dev/null and b/7月核算结果/7月考核车辆多源里程汇总_含车型.xlsx differ
diff --git a/7月核算结果/7月考核车辆多源里程汇总_增加车型前.xlsx b/7月核算结果/7月考核车辆多源里程汇总_增加车型前.xlsx
new file mode 100644
index 0000000..7dffd8b
Binary files /dev/null and b/7月核算结果/7月考核车辆多源里程汇总_增加车型前.xlsx differ
diff --git a/7月核算结果/7月考核车辆多源里程汇总_新日规则前.xlsx b/7月核算结果/7月考核车辆多源里程汇总_新日规则前.xlsx
new file mode 100644
index 0000000..d7b3916
Binary files /dev/null and b/7月核算结果/7月考核车辆多源里程汇总_新日规则前.xlsx differ
diff --git a/7月核算结果/离职人员7月奖金处理明细.xlsx b/7月核算结果/离职人员7月奖金处理明细.xlsx
new file mode 100644
index 0000000..f053a9f
Binary files /dev/null and b/7月核算结果/离职人员7月奖金处理明细.xlsx differ
diff --git a/7月核算结果/租赁任务考核_2026年7月.xlsx b/7月核算结果/租赁任务考核_2026年7月.xlsx
new file mode 100644
index 0000000..f937718
Binary files /dev/null and b/7月核算结果/租赁任务考核_2026年7月.xlsx differ
diff --git a/7月核算结果/租赁任务考核_2026年7月_原版备份.xlsx b/7月核算结果/租赁任务考核_2026年7月_原版备份.xlsx
new file mode 100644
index 0000000..a692a4c
Binary files /dev/null and b/7月核算结果/租赁任务考核_2026年7月_原版备份.xlsx differ
diff --git a/7月核算结果/租赁任务考核_2026年7月_干预情况说明.xlsx b/7月核算结果/租赁任务考核_2026年7月_干预情况说明.xlsx
new file mode 100644
index 0000000..7ddd48d
Binary files /dev/null and b/7月核算结果/租赁任务考核_2026年7月_干预情况说明.xlsx differ
diff --git a/7月核算结果/里程任务考核_7月核算.xlsx b/7月核算结果/里程任务考核_7月核算.xlsx
new file mode 100644
index 0000000..6780d7b
Binary files /dev/null and b/7月核算结果/里程任务考核_7月核算.xlsx differ
diff --git a/7月核算结果/里程任务考核_7月核算_旧规则备份_20260812_183423.xlsx b/7月核算结果/里程任务考核_7月核算_旧规则备份_20260812_183423.xlsx
new file mode 100644
index 0000000..6c3d52a
Binary files /dev/null and b/7月核算结果/里程任务考核_7月核算_旧规则备份_20260812_183423.xlsx differ
diff --git a/7月绩效重算_多源考核里程_20260817/离职人员7月奖金处理明细.xlsx b/7月绩效重算_多源考核里程_20260817/离职人员7月奖金处理明细.xlsx
new file mode 100644
index 0000000..10367a4
Binary files /dev/null and b/7月绩效重算_多源考核里程_20260817/离职人员7月奖金处理明细.xlsx differ
diff --git a/7月绩效重算_多源考核里程_20260817/租赁任务考核_2026年7月.xlsx b/7月绩效重算_多源考核里程_20260817/租赁任务考核_2026年7月.xlsx
new file mode 100644
index 0000000..7feb3bc
Binary files /dev/null and b/7月绩效重算_多源考核里程_20260817/租赁任务考核_2026年7月.xlsx differ
diff --git a/7月绩效重算_多源考核里程_20260817/里程任务考核_7月核算.xlsx b/7月绩效重算_多源考核里程_20260817/里程任务考核_7月核算.xlsx
new file mode 100644
index 0000000..f458c8d
Binary files /dev/null and b/7月绩效重算_多源考核里程_20260817/里程任务考核_7月核算.xlsx differ
diff --git a/_fix_may_flip.py b/_fix_may_flip.py
new file mode 100644
index 0000000..69037db
--- /dev/null
+++ b/_fix_may_flip.py
@@ -0,0 +1,164 @@
+import openpyxl
+from openpyxl.styles import Font, PatternFill, Alignment, Border, Side
+from openpyxl.utils import get_column_letter
+from collections import defaultdict
+
+FP = '/Users/kkfluous/Downloads/租赁任务考核_2026年5月.xlsx'
+
+def num(x):
+    try: return float(x or 0)
+    except: return 0.0
+
+# 读数据
+wb_r = openpyxl.load_workbook(FP, data_only=True)
+ws_r = wb_r['业务考核视图']
+h = [c.value for c in next(ws_r.iter_rows(min_row=1, max_row=1))]
+recs = [dict(zip(h, row)) for row in ws_r.iter_rows(min_row=2, values_only=True) if row[0]]
+
+# 备份仪表盘
+import os
+bak = '/Users/kkfluous/Downloads/租赁任务考核_2026年5月_原版备份.xlsx'
+bak_map = {}
+if os.path.exists(bak):
+    wb_b = openpyxl.load_workbook(bak, data_only=True)
+    for row in wb_b['业务考核视图'].iter_rows(min_row=2, values_only=True):
+        p = str(row[0]).strip() if row[0] else ''
+        bak_map[p] = num(row[14])
+    wb_b.close()
+
+test_recs = [r for r in recs if num(r.get('测试里程(km)')) > 0]
+
+flipped = []
+for r in test_recs:
+    p = str(r['车牌号']).strip()
+    old_km = bak_map.get(p, 0)
+    new_km = num(r.get('实际行驶里程(km)'))
+    t = num(r.get('应考核里程(km)'))
+    if t <= 0: continue
+    if old_km >= t and new_km < t:
+        flipped.append(r)
+
+# 按车型+部门+销售分组
+by_target = defaultdict(lambda: defaultdict(lambda: defaultdict(int)))
+for r in flipped:
+    tgt = r.get('考核目标', '未知车型')
+    dept = r.get('部门名称', '')
+    sales = r.get('销售经理', '')
+    by_target[tgt][dept][sales] += 1
+
+wb_r.close()
+
+# 写sheet
+wb = openpyxl.load_workbook(FP)
+ws = wb['测试与考核情况说明']
+
+SUB   = PatternFill('solid', fgColor='D9E1F2')
+HEAD  = PatternFill('solid', fgColor='4472C4')
+ALT   = PatternFill('solid', fgColor='F2F2F2')
+THIN  = Side('thin', color='B4B4B4')
+BD    = Border(THIN, THIN, THIN, THIN)
+CC = Alignment(horizontal='center', vertical='center', wrap_text=True)
+LL = Alignment(horizontal='left', vertical='center', wrap_text=True)
+
+# 找到section 二所在的行
+section2_row = None
+for r in range(1, ws.max_row + 1):
+    v = ws.cell(row=r, column=1).value
+    if v and '二、测试车辆考核达标情况说明' in str(v):
+        section2_row = r
+        break
+
+if section2_row:
+    # 删除section 二及之后的所有行（保留数据重新写）
+    # 找到从section2_row+1开始到section三之前的所有行
+    section3_row = None
+    for r in range(section2_row + 1, ws.max_row + 1):
+        v = ws.cell(row=r, column=1).value
+        if v and '三、里程' in str(v):
+            section3_row = r
+            break
+    if not section3_row:
+        section3_row = ws.max_row + 1
+
+    # 清除section2+1 到 section3-1 的内容
+    for r in range(section2_row + 1, section3_row):
+        for c in range(1, 7):
+            ws.cell(row=r, column=c).value = None
+            ws.cell(row=r, column=c).fill = PatternFill()
+
+    r = section2_row + 1
+
+    # 描述行
+    ws.merge_cells(start_row=r, start_column=1, end_row=r, end_column=6)
+    ws.cell(row=r, column=1,
+        value=f'共{len(test_recs)}条考核车辆涉及测试；扣减其测试里程后，共 {len(flipped)} 条考核车辆记录由"达标"转判为"未达标"。')
+    ws.cell(row=r, column=1).font = Font(size=10); ws.cell(row=r, column=1).alignment = LL
+    ws.row_dimensions[r].height = 28
+    r += 2
+
+    # 按车型展示，多车型一行
+    targets_sorted = sorted(by_target.keys())
+    triples = [targets_sorted[i:i+2] for i in range(0, len(targets_sorted), 2)]
+
+    for target_group in triples:
+        for ti, tgt in enumerate(target_group):
+            dept_data = by_target[tgt]
+            dept_order = sorted(dept_data.keys(), key=lambda d: -sum(dept_data[d].values()))
+            total_flip = sum(sum(s.values()) for s in dept_data.values())
+
+            base_col = ti * 3 + 1  # 每车型占3列(车型名|部门-销售|数量)
+
+            # 车型标题行
+            ws.merge_cells(start_row=r, start_column=base_col, end_row=r, end_column=base_col+2)
+            ws.cell(row=r, column=base_col, value=f'{tgt}（{total_flip}辆）').fill = SUB
+            ws.cell(row=r, column=base_col).font = Font(bold=True, color='1F4E78')
+            ws.cell(row=r, column=base_col).alignment = CC
+            for j in range(base_col, base_col+3):
+                ws.cell(row=r, column=j).border = BD
+            r += 1
+
+            # 部门-销售-数量表头
+            for j, hh in enumerate(['部门', '销售经理', '数量'], base_col):
+                ws.cell(row=r, column=j, value=hh).fill = HEAD
+                ws.cell(row=r, column=j).font = Font(bold=True, color='FFFFFF')
+                ws.cell(row=r, column=j).alignment = CC
+                ws.cell(row=r, column=j).border = BD
+            r += 1
+
+            # 数据行
+            row_idx = 0
+            for dept in dept_order:
+                for sales in sorted(dept_data[dept].keys()):
+                    cnt = dept_data[dept][sales]
+                    ws.cell(row=r, column=base_col, value=dept).alignment = CC
+                    ws.cell(row=r, column=base_col+1, value=sales).alignment = CC
+                    ws.cell(row=r, column=base_col+2, value=cnt).alignment = CC
+                    for j in range(base_col, base_col+3):
+                        ws.cell(row=r, column=j).border = BD
+                        if row_idx % 2 == 1:
+                            ws.cell(row=r, column=j).fill = ALT
+                    ws.row_dimensions[r].height = 22
+                    r += 1
+                    row_idx += 1
+            # 小计行
+            ws.cell(row=r, column=base_col, value='合计').font = Font(bold=True)
+            ws.cell(row=r, column=base_col).alignment = CC
+            ws.cell(row=r, column=base_col+2, value=total_flip).font = Font(bold=True)
+            ws.cell(row=r, column=base_col+2).alignment = CC
+            for j in range(base_col, base_col+3):
+                ws.cell(row=r, column=j).fill = SUB
+                ws.cell(row=r, column=j).border = BD
+            r += 2
+
+        # 跳到下一个triple的起始行
+        # Find max row reached in this triplet
+        pass  # r already at right position
+
+    r += 1
+
+# 列宽
+for i, w in enumerate([18, 22, 22, 50, 18, 18], 1):
+    ws.column_dimensions[get_column_letter(i)].width = w
+
+wb.save(FP)
+print(f'✅ 更新完成')
diff --git a/_fix_source_may.py b/_fix_source_may.py
new file mode 100644
index 0000000..dc4a0a7
--- /dev/null
+++ b/_fix_source_may.py
@@ -0,0 +1,101 @@
+#!/usr/bin/env python3
+"""Fix May source: remove 4 vehicles from intervention, fix 粤A02239F GPS data."""
+import os, shutil, openpyxl, warnings
+from openpyxl.styles import Font, PatternFill, Alignment, Border, Side
+from openpyxl.utils import get_column_letter
+warnings.filterwarnings('ignore')
+os.chdir('/Users/kkfluous/Downloads')
+
+SRC = '租赁任务考核_2026年5月.xlsx'
+BAK = '租赁任务考核_2026年5月_原版备份.xlsx'
+
+wb = openpyxl.load_workbook(SRC)
+ws = wb['业务考核视图']
+h = [c.value for c in next(ws.iter_rows(min_row=1, max_row=1))]
+print(f"Headers ({len(h)}):")
+for i, hh in enumerate(h): print(f"  [{i}] {hh}")
+
+# Column indices
+COL_PLATE = 0
+COL_ACTUAL = h.index('实际行驶里程(km)')  # GPS or仪表盘
+COL_TBOX = h.index('原TBOX里程(km)')
+COL_TEST = h.index('测试里程(km)')
+COL_RATE = h.index('完成率(%)')
+COL_PASS = h.index('是否达标')
+
+YELLOW = PatternFill('solid', fgColor='FFFF00')
+NO_FILL = PatternFill()  # default/empty
+
+# ── Fix 1: Remove 4 vehicles from intervention ──
+REMOVE = {'粤A03186F', '粤A01128F', '粤A08391F', '粤A01396F'}
+
+removed_count = 0
+for row in ws.iter_rows(min_row=2, max_row=ws.max_row):
+    plate = str(row[COL_PLATE].value or '').strip()
+    if plate not in REMOVE:
+        continue
+
+    tbox = row[COL_TBOX].value  # this is the original仪表盘里程
+    if tbox is None or float(tbox or 0) <= 0:
+        continue
+
+    original_km = float(tbox)
+    target_km = float(row[h.index('应考核里程(km)')].value or 0)
+
+    # Restore: 实际行驶里程 = original仪表盘
+    row[COL_ACTUAL].value = original_km
+    row[COL_TBOX].value = None  # clear TBOX (non-intervention)
+    row[COL_TEST].value = 0     # clear test
+
+    # Recalculate
+    rate = round(original_km / target_km * 100, 2) if target_km > 0 else 0
+    new_pass = '达标' if original_km >= target_km and target_km > 0 else '未达标'
+    row[COL_RATE].value = rate
+    row[COL_PASS].value = new_pass
+
+    # Remove yellow highlight
+    for c in range(len(h)):
+        row[c].fill = NO_FILL
+
+    removed_count += 1
+    print(f"  移除: {plate} 实际恢复={original_km} 达标={new_pass}")
+
+print(f"移除干预: {removed_count} 辆")
+
+# ── Fix 2: 粤A02239F GPS = 539.7 ──
+plate = '粤A02239F'
+for row in ws.iter_rows(min_row=2, max_row=ws.max_row):
+    p = str(row[COL_PLATE].value or '').strip()
+    if p != plate:
+        continue
+
+    tbox = float(row[COL_TBOX].value or 0)
+    gps_new = 539.7
+    target_km = float(row[h.index('应考核里程(km)')].value or 0)
+
+    row[COL_ACTUAL].value = gps_new
+    row[COL_TEST].value = round(tbox - gps_new, 2)
+    rate = round(gps_new / target_km * 100, 2) if target_km > 0 else 0
+    new_pass = '达标' if gps_new >= target_km and target_km > 0 else '未达标'
+    row[COL_RATE].value = rate
+    row[COL_PASS].value = new_pass
+
+    print(f"\n修复: {plate} GPS={gps_new} TBOX={tbox} Test={round(tbox-gps_new,2)} 达标={new_pass}")
+    break
+
+# ── Save ──
+wb.save(SRC)
+print(f"\n✅ 已保存: {SRC}")
+
+# ── Quick stats ──
+wb2 = openpyxl.load_workbook(SRC, data_only=True)
+ws2 = wb2['业务考核视图']
+h2 = [c.value for c in next(ws2.iter_rows(min_row=1, max_row=1))]
+total = 0; pass_count = 0
+for row in ws2.iter_rows(min_row=2, values_only=True):
+    if not row[0]: continue
+    r = dict(zip(h2, row))
+    total += 1
+    if r.get('是否达标') == '达标': pass_count += 1
+wb2.close()
+print(f"总记录={total} 达标={pass_count}")
diff --git a/_gen_intervention_may.py b/_gen_intervention_may.py
new file mode 100644
index 0000000..60054a6
--- /dev/null
+++ b/_gen_intervention_may.py
@@ -0,0 +1,327 @@
+#!/usr/bin/env python3
+"""Generate 租赁任务考核_2026年5月_干预情况说明.xlsx based on updated source data."""
+import os, openpyxl
+from openpyxl.styles import Font, PatternFill, Alignment, Border, Side
+from openpyxl.utils import get_column_letter
+from collections import defaultdict
+from copy import copy
+
+os.chdir('/Users/kkfluous/Downloads')
+
+SRC = '租赁任务考核_2026年5月.xlsx'
+BAK = '租赁任务考核_2026年5月_原版备份.xlsx'
+
+# ── Read source ──
+wb_s = openpyxl.load_workbook(SRC, data_only=True)
+ws_s = wb_s['业务考核视图']
+src_headers = [c.value for c in next(ws_s.iter_rows(min_row=1, max_row=1))]
+src_data = list(ws_s.iter_rows(min_row=2, values_only=True))
+wb_s.close()
+
+# ── Read backup ──
+wb_b = openpyxl.load_workbook(BAK, data_only=True)
+ws_b = wb_b['业务考核视图']
+bak_data = list(ws_b.iter_rows(min_row=2, values_only=True))
+wb_b.close()
+
+print(f"Source: {len(src_data)} records")
+print(f"Backup: {len(bak_data)} records")
+
+# ── Compare ──
+# Build plate -> row map for backup (keep all, including duplicates)
+def make_key(row):
+    plate = str(row[0]).strip() if row[0] else ''
+    sales = str(row[4]).strip() if len(row) > 4 and row[4] else ''
+    target = str(row[9]).strip() if len(row) > 9 and row[9] else ''
+    return (plate, sales, target)
+
+bak_map = {}
+for r in bak_data:
+    k = make_key(r)
+    bak_map[k] = r
+
+src_map = {}
+for r in src_data:
+    k = make_key(r)
+    src_map[k] = r
+
+# Find new/changed/removed
+new_records = []
+changed_records = []
+removed_records = []
+
+for k, sr in src_map.items():
+    if k not in bak_map:
+        new_records.append(sr)
+    else:
+        br = bak_map[k]
+        old_pass = str(br[16]).strip() if len(br) > 16 else ''
+        new_pass = str(sr[16]).strip() if len(sr) > 16 else ''
+        old_km = float(br[14] or 0) if len(br) > 14 else 0
+        new_km = float(sr[14] or 0) if len(sr) > 14 else 0
+        old_rate = float(br[15] or 0) if len(br) > 15 else 0
+        new_rate = float(sr[15] or 0) if len(sr) > 15 else 0
+        if old_pass != new_pass or abs(old_km - new_km) > 0.01 or abs(old_rate - new_rate) > 0.01:
+            changed_records.append((br, sr))
+
+for k, br in bak_map.items():
+    if k not in src_map:
+        removed_records.append(br)
+
+print(f"New: {len(new_records)}, Changed: {len(changed_records)}, Removed: {len(removed_records)}")
+
+# Count达标 changes
+old_pass_count = sum(1 for r in bak_data if str(r[16]).strip() == '达标' if len(r) > 16)
+new_pass_count = sum(1 for r in src_data if str(r[16]).strip() == '达标' if len(r) > 16)
+print(f"达标: {old_pass_count} -> {new_pass_count}")
+
+# Flip analysis (达标 → 未达标)
+flipped = []
+for old_r, new_r in changed_records:
+    old_p = str(old_r[16]).strip() if len(old_r) > 16 else ''
+    new_p = str(new_r[16]).strip() if len(new_r) > 16 else ''
+    if old_p == '达标' and new_p == '未达标':
+        flipped.append((old_r, new_r))
+# Reverse flip (未达标 → 达标)
+reversed_flip = []
+for old_r, new_r in changed_records:
+    old_p = str(old_r[16]).strip() if len(old_r) > 16 else ''
+    new_p = str(new_r[16]).strip() if len(new_r) > 16 else ''
+    if old_p == '未达标' and new_p == '达标':
+        reversed_flip.append((old_r, new_r))
+
+print(f"Flipped (达标→未达标): {len(flipped)}")
+print(f"Reversed (未达标→达标): {len(reversed_flip)}")
+
+# ── Generate Excel ──
+wb = openpyxl.Workbook()
+
+# Styles
+TITLE_FONT = Font(bold=True, size=14)
+SECTION_FONT = Font(bold=True, size=12, color='1F4E78')
+HEADER_FILL = PatternFill('solid', fgColor='4472C4')
+HEADER_FONT = Font(bold=True, color='FFFFFF', size=10)
+SUB_FILL = PatternFill('solid', fgColor='D9E1F2')
+ALT_FILL = PatternFill('solid', fgColor='F2F2F2')
+NOTE_FONT = Font(size=10)
+BOLD_FONT = Font(bold=True, size=10)
+THIN = Side('thin', color='B4B4B4')
+BORDER = Border(THIN, THIN, THIN, THIN)
+CC = Alignment(horizontal='center', vertical='center', wrap_text=True)
+LL = Alignment(horizontal='left', vertical='center', wrap_text=True)
+NUM_FMT = '#,##0.00'
+
+def style_header_row(ws, row, ncols, fill=HEADER_FILL, font=HEADER_FONT):
+    for c in range(1, ncols + 1):
+        cell = ws.cell(row=row, column=c)
+        cell.fill = fill
+        cell.font = font
+        cell.alignment = CC
+        cell.border = BORDER
+
+def style_data_row(ws, row, ncols, alt=False):
+    for c in range(1, ncols + 1):
+        cell = ws.cell(row=row, column=c)
+        cell.border = BORDER
+        cell.font = Font(size=10)
+        cell.alignment = CC
+        if alt:
+            cell.fill = ALT_FILL
+
+# ── Sheet 1: 测试与考核情况说明 ──
+ws1 = wb.active
+ws1.title = '测试与考核情况说明'
+ws1.sheet_properties.tabColor = '4472C4'
+
+ws1.merge_cells('A1:G1')
+ws1.cell(row=1, column=1, value='2026年5月 业务考核 — 数据更新说明').font = TITLE_FONT
+ws1.cell(row=1, column=1).alignment = CC
+ws1.row_dimensions[1].height = 30
+
+ws1.merge_cells('A2:G2')
+desc = (
+    '5月业务考核源数据已于近期更新。'
+    f'共{len(src_data)}条考核记录，{new_pass_count}条达标（原{old_pass_count}条）。\n'
+    f'本次数据更新：新增{len(new_records)}条，变更{len(changed_records)}条，删除{len(removed_records)}条。\n'
+    f'其中达标→未达标翻转{len(flipped)}条，未达标→达标翻转{len(reversed_flip)}条。\n'
+    '本文件记录更新前后的数据对比，供核查参考。'
+)
+ws1.cell(row=2, column=1, value=desc).font = NOTE_FONT
+ws1.cell(row=2, column=1).alignment = Alignment(horizontal='left', vertical='top', wrap_text=True)
+ws1.row_dimensions[2].height = 80
+
+r = 4
+ws1.merge_cells(start_row=r, start_column=1, end_row=r, end_column=7)
+ws1.cell(row=r, column=1, value='一、5月业务考核概况').font = SECTION_FONT
+ws1.cell(row=r, column=1).alignment = LL
+r += 1
+
+summary_data = [
+    ('项目', '5月业务考核车辆', '变更记录', '占比/变化'),
+    ('考核记录数(条)', len(src_data), len(changed_records), f'{len(changed_records)/len(src_data)*100:.1f}%' if len(src_data) > 0 else '0%'),
+    ('达标记录数(条)', new_pass_count, new_pass_count - old_pass_count, f'{"+" if new_pass_count >= old_pass_count else ""}{new_pass_count - old_pass_count}'),
+    ('未达标记录数(条)', len(src_data) - new_pass_count, (len(src_data) - new_pass_count) - (len(bak_data) - old_pass_count), ''),
+    ('达标率', f'{new_pass_count/len(src_data)*100:.1f}%' if len(src_data) > 0 else '0%', '', ''),
+]
+
+for i, row_data in enumerate(summary_data):
+    for c, val in enumerate(row_data, 1):
+        ws1.cell(row=r, column=c, value=val)
+    if i == 0:
+        style_header_row(ws1, r, 4)
+    else:
+        style_data_row(ws1, r, 4, alt=(i % 2 == 0))
+        ws1.cell(row=r, column=1).alignment = LL
+    r += 1
+
+r += 1
+ws1.merge_cells(start_row=r, start_column=1, end_row=r, end_column=7)
+ws1.cell(row=r, column=1, value='二、考核记录变更明细').font = SECTION_FONT
+ws1.cell(row=r, column=1).alignment = LL
+r += 1
+
+if flipped:
+    ws1.merge_cells(start_row=r, start_column=1, end_row=r, end_column=7)
+    ws1.cell(row=r, column=1, value=f'2.1 达→未达标翻转（{len(flipped)}条）').font = Font(bold=True, color='C00000')
+    r += 1
+
+    flip_headers = ['车牌号', '部门', '销售经理', '客户名称', '考核目标', '原实际里程', '新实际里程', '原是否达标', '新是否达标']
+    for c, h in enumerate(flip_headers, 1):
+        ws1.cell(row=r, column=c, value=h)
+    style_header_row(ws1, r, len(flip_headers))
+    r += 1
+
+    for i, (old_r, new_r) in enumerate(flipped):
+        vals = [
+            str(old_r[0]) if old_r[0] else '',
+            str(old_r[3]) if len(old_r) > 3 else '',
+            str(old_r[4]) if len(old_r) > 4 else '',
+            str(old_r[5]) if len(old_r) > 5 else '',
+            str(old_r[9]) if len(old_r) > 9 else '',
+            round(float(old_r[14] or 0), 2) if len(old_r) > 14 and old_r[14] else 0,
+            round(float(new_r[14] or 0), 2) if len(new_r) > 14 and new_r[14] else 0,
+            str(old_r[16]) if len(old_r) > 16 else '',
+            str(new_r[16]) if len(new_r) > 16 else '',
+        ]
+        for c, val in enumerate(vals, 1):
+            ws1.cell(row=r, column=c, value=val)
+        style_data_row(ws1, r, len(flip_headers), alt=(i % 2 == 1))
+        r += 1
+    r += 1
+
+if reversed_flip:
+    ws1.merge_cells(start_row=r, start_column=1, end_row=r, end_column=7)
+    ws1.cell(row=r, column=1, value=f'2.2 未达标→达标翻转（{len(reversed_flip)}条）').font = Font(bold=True, color='008000')
+    r += 1
+
+    for c, h in enumerate(flip_headers, 1):
+        ws1.cell(row=r, column=c, value=h)
+    style_header_row(ws1, r, len(flip_headers))
+    r += 1
+
+    for i, (old_r, new_r) in enumerate(reversed_flip):
+        vals = [
+            str(old_r[0]) if old_r[0] else '',
+            str(old_r[3]) if len(old_r) > 3 else '',
+            str(old_r[4]) if len(old_r) > 4 else '',
+            str(old_r[5]) if len(old_r) > 5 else '',
+            str(old_r[9]) if len(old_r) > 9 else '',
+            round(float(old_r[14] or 0), 2) if len(old_r) > 14 and old_r[14] else 0,
+            round(float(new_r[14] or 0), 2) if len(new_r) > 14 and new_r[14] else 0,
+            str(old_r[16]) if len(old_r) > 16 else '',
+            str(new_r[16]) if len(new_r) > 16 else '',
+        ]
+        for c, val in enumerate(vals, 1):
+            ws1.cell(row=r, column=c, value=val)
+        style_data_row(ws1, r, len(flip_headers), alt=(i % 2 == 1))
+        r += 1
+    r += 1
+
+if other_changes := [(o, n) for o, n in changed_records
+                      if not (str(o[16]).strip() == '达标' and str(n[16]).strip() == '未达标')
+                      and not (str(o[16]).strip() == '未达标' and str(n[16]).strip() == '达标')]:
+    ws1.merge_cells(start_row=r, start_column=1, end_row=r, end_column=7)
+    ws1.cell(row=r, column=1, value=f'2.3 其他变更（{len(other_changes)}条，里程变化但达标状态不变）').font = BOLD_FONT
+    r += 1
+
+col_widths = [14, 12, 12, 30, 22, 16, 16, 12, 12]
+for i, w in enumerate(col_widths, 1):
+    ws1.column_dimensions[get_column_letter(i)].width = w
+
+# ── Sheet 2: 业务考核视图 (current data) ──
+ws2 = wb.create_sheet('业务考核视图')
+ws2.sheet_properties.tabColor = '70AD47'
+
+for c, h in enumerate(src_headers, 1):
+    ws2.cell(row=1, column=c, value=h)
+style_header_row(ws2, 1, len(src_headers))
+
+for i, row in enumerate(src_data):
+    for c, val in enumerate(row, 1):
+        ws2.cell(row=i + 2, column=c, value=val)
+    style_data_row(ws2, i + 2, len(src_headers), alt=(i % 2 == 1))
+
+for i, w in enumerate([12, 8, 8, 12, 10, 28, 20, 24, 10, 22, 12, 12, 10, 16, 16, 10, 10, 10, 12, 12, 12], 1):
+    ws2.column_dimensions[get_column_letter(i)].width = w
+
+# ── Sheet 3: 5月业务考核车辆总览 ──
+ws3 = wb.create_sheet('5月业务考核车辆总览')
+ws3.sheet_properties.tabColor = 'ED7D31'
+
+for c, h in enumerate(src_headers, 1):
+    ws3.cell(row=1, column=c, value=h)
+style_header_row(ws3, 1, len(src_headers))
+
+for i, row in enumerate(src_data):
+    for c, val in enumerate(row, 1):
+        ws3.cell(row=i + 2, column=c, value=val)
+    style_data_row(ws3, i + 2, len(src_headers), alt=(i % 2 == 1))
+
+for i, w in enumerate([12, 8, 8, 12, 10, 28, 20, 24, 10, 22, 12, 12, 10, 16, 16, 10, 10, 10, 12, 12, 12], 1):
+    ws3.column_dimensions[get_column_letter(i)].width = w
+
+# ── Sheet 4: 涉及变更的业务考核车辆 ──
+ws4 = wb.create_sheet('涉及变更的业务考核车辆')
+ws4.sheet_properties.tabColor = 'FFC000'
+
+ws4.merge_cells('A1:U1')
+ws4.cell(row=1, column=1, value=f'变更记录汇总 — 新增{len(new_records)}条 / 变更{len(changed_records)}条 / 删除{len(removed_records)}条').font = Font(bold=True, size=11)
+ws4.cell(row=1, column=1).alignment = CC
+ws4.row_dimensions[1].height = 24
+
+# Source headers + change type col
+ext_headers = list(src_headers) + ['变更类型', '原实际里程', '原是否达标']
+for c, h in enumerate(ext_headers, 1):
+    ws4.cell(row=3, column=c, value=h)
+style_header_row(ws4, 3, len(ext_headers))
+
+r = 4
+# New records
+for i, rec in enumerate(new_records):
+    for c, val in enumerate(rec, 1):
+        ws4.cell(row=r, column=c, value=val)
+    ws4.cell(row=r, column=len(src_headers) + 1, value='新增')
+    style_data_row(ws4, r, len(ext_headers), alt=(i % 2 == 1))
+    ws4.row_dimensions[r].height = 20
+    r += 1
+
+# Changed records
+for i, (old_r, new_r) in enumerate(changed_records):
+    for c, val in enumerate(new_r, 1):
+        ws4.cell(row=r, column=c, value=val)
+    ws4.cell(row=r, column=len(src_headers) + 1, value='变更')
+    old_km = round(float(old_r[14] or 0), 2) if len(old_r) > 14 and old_r[14] else 0
+    old_pass = str(old_r[16]) if len(old_r) > 16 else ''
+    ws4.cell(row=r, column=len(src_headers) + 2, value=old_km)
+    ws4.cell(row=r, column=len(src_headers) + 3, value=old_pass)
+    style_data_row(ws4, r, len(ext_headers), alt=(i % 2 == 1))
+    ws4.row_dimensions[r].height = 20
+    r += 1
+
+for i, w in enumerate([12, 8, 8, 12, 10, 28, 20, 24, 10, 22, 12, 12, 10, 16, 16, 10, 10, 10, 12, 12, 12, 10, 14, 12], 1):
+    ws4.column_dimensions[get_column_letter(i)].width = w
+
+# ── Save ──
+out_path = 'new-kpi-202605_1/租赁任务考核_2026年5月_干预情况说明.xlsx'
+wb.save(out_path)
+print(f'\n✅ {out_path} ({len(wb.sheetnames)} sheets)')
diff --git a/_gen_intervention_v2.py b/_gen_intervention_v2.py
new file mode 100644
index 0000000..8438de1
--- /dev/null
+++ b/_gen_intervention_v2.py
@@ -0,0 +1,330 @@
+#!/usr/bin/env python3
+"""Generate 租赁任务考核_2026年5月_干预情况说明.xlsx — GPS干预后的完整说明"""
+import os, openpyxl
+from openpyxl.styles import Font, PatternFill, Alignment, Border, Side
+from openpyxl.utils import get_column_letter
+from collections import defaultdict
+
+os.chdir('/Users/kkfluous/Downloads')
+
+SRC = '租赁任务考核_2026年5月.xlsx'
+BAK = '租赁任务考核_2026年5月_原版备份.xlsx'
+INTV = '标黄为5月干预车辆.xlsx'
+
+# Styles
+HEAD = PatternFill('solid', fgColor='4472C4')
+HFONT = Font(bold=True, color='FFFFFF', size=10)
+SUB = PatternFill('solid', fgColor='D9E1F2')
+ALT = PatternFill('solid', fgColor='F2F2F2')
+TITLE_FONT = Font(bold=True, size=14)
+SEC_FONT = Font(bold=True, size=12, color='1F4E78')
+NOTE_FONT = Font(size=10)
+BOLD_FONT = Font(bold=True, size=10)
+RED_FONT = Font(size=10, color='C00000')
+BORD = Border(Side('thin','B4B4B4'), Side('thin','B4B4B4'), Side('thin','B4B4B4'), Side('thin','B4B4B4'))
+CC = Alignment(horizontal='center', vertical='center', wrap_text=True)
+LL = Alignment(horizontal='left', vertical='center', wrap_text=True)
+
+# ── Load intervention plates ──
+wb_i = openpyxl.load_workbook(INTV)
+ws_i = wb_i['Sheet1']
+yellow_plates = set()
+for row in ws_i.iter_rows(min_row=2, values_only=False):
+    cell = row[0]
+    fill = cell.fill
+    is_yellow = (fill and fill.patternType == 'solid' and fill.start_color
+                 and fill.start_color.rgb and 'FFFF00' in str(fill.start_color.rgb))
+    if is_yellow:
+        p = str(cell.value).strip() if cell.value else ''
+        if p:
+            yellow_plates.add(p)
+wb_i.close()
+
+# ── Load source (post-intervention) and backup (pre-intervention) ──
+def load_records(fp):
+    wb = openpyxl.load_workbook(fp, data_only=True)
+    ws = wb['业务考核视图']
+    h = [c.value for c in next(ws.iter_rows(min_row=1, max_row=1))]
+    recs = []
+    for row in ws.iter_rows(min_row=2, values_only=True):
+        if row[0]:
+            recs.append(dict(zip(h, row)))
+    wb.close()
+    return recs
+
+src_recs = load_records(SRC)
+bak_recs = load_records(BAK)
+
+# Build plate -> {backup rec, source rec}
+bak_by_plate = {}
+for r in bak_recs:
+    p = str(r['车牌号']).strip()
+    # For duplicate plates, keep last
+    key = (p, str(r.get('销售经理','')), str(r.get('项目名称','')))
+    bak_by_plate[key] = r
+
+src_by_plate = {}
+for r in src_recs:
+    p = str(r['车牌号']).strip()
+    key = (p, str(r.get('销售经理','')), str(r.get('项目名称','')))
+    src_by_plate[key] = r
+
+# ── Analyze intervention impact ──
+# For each source record that's an intervention vehicle, find its backup
+intv_records = []
+for key, sr in src_by_plate.items():
+    p = key[0]
+    if p not in yellow_plates:
+        continue
+    br = bak_by_plate.get(key)
+    tbox = sr.get('原TBOX里程(km)')
+    gps = sr.get('实际行驶里程(km)')
+    test = sr.get('测试里程(km)')
+    old_actual = float(br['实际行驶里程(km)'] or 0) if br else 0
+    old_pass = str(br['是否达标']).strip() if br else '?'
+    new_pass = str(sr['是否达标']).strip() if sr else '?'
+    target = float(sr.get('应考核里程(km)', 0) or 0)
+
+    tbox_val = float(tbox or 0)
+    gps_val = float(gps or 0)
+    test_val = float(test or 0)
+
+    flip = ''
+    if old_pass == '达标' and new_pass == '未达标':
+        flip = '达→未达标'
+    elif old_pass == '未达标' and new_pass == '达标':
+        flip = '未→达标'
+
+    has_gps = '有' if gps_val > 0 else '无GPS数据'
+
+    intv_records.append({
+        'plate': p, 'dept': sr.get('部门名称', ''), 'sales': sr.get('销售经理', ''),
+        'client': sr.get('客户名称', ''), 'target_name': sr.get('考核目标', ''),
+        'target_km': target, 'old_actual': old_actual, 'tbox': tbox_val,
+        'gps': gps_val, 'test': test_val, 'old_pass': old_pass, 'new_pass': new_pass,
+        'flip': flip, 'has_gps': has_gps
+    })
+
+print(f"Intervention records: {len(intv_records)}")
+flip_down = [r for r in intv_records if r['flip'] == '达→未达标']
+flip_up = [r for r in intv_records if r['flip'] == '未→达标']
+no_gps = [r for r in intv_records if r['gps'] == 0]
+print(f"达→未达标: {len(flip_down)}, 未→达标: {len(flip_up)}, 无GPS: {len(no_gps)}")
+
+# ── Intervention vehicles by target type ──
+by_target = defaultdict(list)
+for r in intv_records:
+    by_target[r['target_name']].append(r)
+
+# ── Generate workbook ──
+wb = openpyxl.Workbook()
+
+def style_header(ws, row, ncols):
+    for c in range(1, ncols+1):
+        cell = ws.cell(row=row, column=c)
+        cell.fill = HEAD; cell.font = HFONT; cell.alignment = CC; cell.border = BORD
+
+def style_row(ws, row, ncols, alt=False):
+    for c in range(1, ncols+1):
+        cell = ws.cell(row=row, column=c)
+        cell.border = BORD; cell.font = Font(size=10); cell.alignment = CC
+        if alt: cell.fill = ALT
+
+# ── Sheet 1: 测试与考核情况说明 ──
+ws1 = wb.active
+ws1.title = '测试与考核情况说明'
+ws1.sheet_properties.tabColor = '4472C4'
+
+ws1.merge_cells('A1:H1')
+ws1.cell(row=1, column=1, value='2026年5月 业务考核 — GPS干预与考核情况说明').font = TITLE_FONT
+ws1.cell(row=1, column=1).alignment = CC
+ws1.row_dimensions[1].height = 32
+
+desc = (
+    f'5月共计{len(src_recs)}条考核记录，其中{len(yellow_plates)}辆车/{len(intv_records)}条记录涉及GPS里程干预。\n'
+    '干预规则：对于标黄车辆，使用G7平台GPS每日里程（考核期内逐日累加）替代原仪表盘里程作为实际行驶里程参与考核。\n'
+    f'GPS数据来源：hydrogen_energy.ln_vehicle_g7_mileage（101.133.130.65），按考核起止时间汇总每日total_mileage/100000=km。\n'
+    f'干预结果：{len(flip_down)}条由达标翻转为未达标，{len(flip_up)}条由未达标翻转为达标，{len(no_gps)}条车辆无GPS数据（里程记为0）。'
+)
+ws1.merge_cells('A2:H2')
+ws1.cell(row=2, column=1, value=desc).font = NOTE_FONT
+ws1.cell(row=2, column=1).alignment = Alignment(horizontal='left', vertical='top', wrap_text=True)
+ws1.row_dimensions[2].height = 80
+
+r = 4
+ws1.merge_cells(start_row=r, start_column=1, end_row=r, end_column=8)
+ws1.cell(row=r, column=1, value='一、5月考核概况（GPS干预后）').font = SEC_FONT
+r += 1
+
+# Summary table
+summary_headers = ['指标', '干预前', '干预后', '变化']
+for c, h in enumerate(summary_headers, 1):
+    ws1.cell(row=r, column=c, value=h)
+style_header(ws1, r, 4)
+r += 1
+
+old_pass_count = sum(1 for r in bak_recs if str(r.get('是否达标','')).strip() == '达标')
+new_pass_count = sum(1 for r in src_recs if str(r.get('是否达标','')).strip() == '达标')
+old_fail = len(bak_recs) - old_pass_count
+new_fail = len(src_recs) - new_pass_count
+
+summary_rows = [
+    ('考核记录数(条)', len(bak_recs), len(src_recs), ''),
+    ('达标记录数(条)', old_pass_count, new_pass_count, f'{new_pass_count - old_pass_count:+d}'),
+    ('未达标记录数(条)', old_fail, new_fail, f'{new_fail - old_fail:+d}'),
+    ('达标率', f'{old_pass_count/len(bak_recs)*100:.1f}%', f'{new_pass_count/len(src_recs)*100:.1f}%', ''),
+    ('干预车辆数', len(yellow_plates), len(yellow_plates), ''),
+    ('涉及考核记录', len(intv_records), len(intv_records), ''),
+    ('有GPS数据', '', sum(1 for r in intv_records if r['gps'] > 0), ''),
+    ('无GPS数据(里程=0)', '', len(no_gps), ''),
+]
+
+for row_data in summary_rows:
+    for c, val in enumerate(row_data, 1):
+        ws1.cell(row=r, column=c, value=val)
+    style_row(ws1, r, 4)
+    r += 1
+
+r += 1
+ws1.merge_cells(start_row=r, start_column=1, end_row=r, end_column=8)
+ws1.cell(row=r, column=1, value='二、干预车型达标翻转统计').font = SEC_FONT
+r += 1
+
+# By target type
+flip_headers = ['考核目标', '干预车辆数', '达→未达标', '未→达标', '有GPS数据', '无GPS数据']
+for c, h in enumerate(flip_headers, 1):
+    ws1.cell(row=r, column=c, value=h)
+style_header(ws1, r, 6)
+r += 1
+
+# Sort: most affected first
+target_order = sorted(by_target.keys(), key=lambda t: -len(by_target[t]))
+for i, tgt in enumerate(target_order):
+    recs = by_target[tgt]
+    ws1.cell(row=r, column=1, value=tgt)
+    ws1.cell(row=r, column=2, value=len(recs))
+    ws1.cell(row=r, column=3, value=sum(1 for x in recs if x['flip'] == '达→未达标'))
+    ws1.cell(row=r, column=4, value=sum(1 for x in recs if x['flip'] == '未→达标'))
+    ws1.cell(row=r, column=5, value=sum(1 for x in recs if x['gps'] > 0))
+    ws1.cell(row=r, column=6, value=sum(1 for x in recs if x['gps'] == 0))
+    style_row(ws1, r, 6, alt=(i%2==1))
+    r += 1
+
+col_widths1 = [14, 15, 15, 15, 18, 18, 18, 18]
+for i, w in enumerate(col_widths1, 1):
+    ws1.column_dimensions[get_column_letter(i)].width = w
+
+# ── Sheet 2: 干预车辆明细 ──
+ws2 = wb.create_sheet('干预车辆明细')
+ws2.sheet_properties.tabColor = 'ED7D31'
+
+ws2.merge_cells('A1:N1')
+ws2.cell(row=1, column=1, value='5月干预车辆 — 原仪表盘里程 vs GPS里程 逐条对比').font = TITLE_FONT
+ws2.cell(row=1, column=1).alignment = CC
+ws2.row_dimensions[1].height = 28
+
+detail_headers = ['车牌号', '部门', '销售经理', '客户名称', '考核目标',
+                   '考核天数', '应考核里程(km)', '原仪表盘里程(km)', 'GPS行驶里程(km)',
+                   '测试里程(km)', '原是否达标', '干预后达标', '翻转', 'GPS数据']
+r2 = 3
+for c, h in enumerate(detail_headers, 1):
+    ws2.cell(row=r2, column=c, value=h)
+style_header(ws2, r2, 14)
+r2 += 1
+
+# Sort: flip-down first, then no-GPS, then others
+intv_records.sort(key=lambda r: (0 if r['flip'] == '达→未达标' else 1 if r['gps'] == 0 else 2, r['plate']))
+
+for i, rec in enumerate(intv_records):
+    ws2.cell(row=r2, column=1, value=rec['plate'])
+    ws2.cell(row=r2, column=2, value=rec['dept'])
+    ws2.cell(row=r2, column=3, value=rec['sales'])
+    ws2.cell(row=r2, column=4, value=rec['client'])
+    ws2.cell(row=r2, column=5, value=rec['target_name'])
+    ws2.cell(row=r2, column=6, value=rec.get('days', ''))
+    ws2.cell(row=r2, column=7, value=round(rec['target_km'], 2))
+    ws2.cell(row=r2, column=8, value=round(rec['tbox'], 2))
+    ws2.cell(row=r2, column=9, value=round(rec['gps'], 2))
+    ws2.cell(row=r2, column=10, value=round(rec['test'], 2))
+    ws2.cell(row=r2, column=11, value=rec['old_pass'])
+    ws2.cell(row=r2, column=12, value=rec['new_pass'])
+    # Flip column with color
+    flip_cell = ws2.cell(row=r2, column=13, value=rec['flip'])
+    if rec['flip'] == '达→未达标':
+        flip_cell.font = RED_FONT
+    ws2.cell(row=r2, column=14, value=rec['has_gps'])
+
+    style_row(ws2, r2, 14, alt=(i%2==1))
+    r2 += 1
+
+col_widths2 = [12, 10, 10, 26, 20, 8, 14, 16, 16, 14, 10, 10, 10, 10]
+for i, w in enumerate(col_widths2, 1):
+    ws2.column_dimensions[get_column_letter(i)].width = w
+
+# ── Sheet 3: 业务考核视图 (current, post-intervention) ──
+ws3 = wb.create_sheet('业务考核视图')
+ws3.sheet_properties.tabColor = '70AD47'
+
+src_h = list(src_recs[0].keys())
+for c, h in enumerate(src_h, 1):
+    ws3.cell(row=1, column=c, value=h)
+style_header(ws3, 1, len(src_h))
+
+for i, rec in enumerate(src_recs):
+    for c, h in enumerate(src_h, 1):
+        val = rec.get(h)
+        if val is not None:
+            ws3.cell(row=i+2, column=c, value=val)
+    style_row(ws3, i+2, len(src_h), alt=(i%2==1))
+
+for i in range(1, len(src_h)+1):
+    ws3.column_dimensions[get_column_letter(i)].width = 12
+
+# ── Sheet 4: 干预车辆汇总(按部门+销售) ──
+ws4 = wb.create_sheet('干预汇总(按部门销售)')
+ws4.sheet_properties.tabColor = 'FFC000'
+
+ws4.merge_cells('A1:F1')
+ws4.cell(row=1, column=1, value='干预车辆 — 按部门+销售经理汇总').font = TITLE_FONT
+ws4.cell(row=1, column=1).alignment = CC
+
+# Aggregate
+by_dept_sales = defaultdict(lambda: {'total': 0, 'flip_down': 0, 'flip_up': 0, 'no_gps': 0, 'has_gps': 0, 'gps_amount': 0})
+for rec in intv_records:
+    key = (rec['dept'], rec['sales'])
+    by_dept_sales[key]['total'] += 1
+    if rec['flip'] == '达→未达标':
+        by_dept_sales[key]['flip_down'] += 1
+    elif rec['flip'] == '未→达标':
+        by_dept_sales[key]['flip_up'] += 1
+    if rec['gps'] > 0:
+        by_dept_sales[key]['has_gps'] += 1
+        by_dept_sales[key]['gps_amount'] += rec['gps']
+    else:
+        by_dept_sales[key]['no_gps'] += 1
+
+agg_headers = ['部门', '销售经理', '干预车辆数', '达→未达标', '未→达标', '无GPS数据']
+r4 = 3
+for c, h in enumerate(agg_headers, 1):
+    ws4.cell(row=r4, column=c, value=h)
+style_header(ws4, r4, 6)
+r4 += 1
+
+for (dept, sales), data in sorted(by_dept_sales.items()):
+    ws4.cell(row=r4, column=1, value=dept)
+    ws4.cell(row=r4, column=2, value=sales)
+    ws4.cell(row=r4, column=3, value=data['total'])
+    ws4.cell(row=r4, column=4, value=data['flip_down'])
+    ws4.cell(row=r4, column=5, value=data['flip_up'])
+    ws4.cell(row=r4, column=6, value=data['no_gps'])
+    style_row(ws4, r4, 6)
+    r4 += 1
+
+col_widths4 = [14, 12, 12, 12, 12, 12]
+for i, w in enumerate(col_widths4, 1):
+    ws4.column_dimensions[get_column_letter(i)].width = w
+
+# ── Save ──
+out_path = 'new-kpi-202605_1/租赁任务考核_2026年5月_干预情况说明.xlsx'
+wb.save(out_path)
+print(f'✅ {out_path} ({len(wb.sheetnames)} sheets)')
diff --git a/_gen_intervention_v3.py b/_gen_intervention_v3.py
new file mode 100644
index 0000000..c6c1e67
--- /dev/null
+++ b/_gen_intervention_v3.py
@@ -0,0 +1,410 @@
+#!/usr/bin/env python3
+"""生成 租赁任务考核_2026年5月_干预情况说明.xlsx — 公式计算 + %格式 + 完整匹配参考样式"""
+import os, openpyxl
+from openpyxl.styles import Font, PatternFill, Alignment, Border, Side, numbers
+from openpyxl.utils import get_column_letter
+from collections import defaultdict
+
+os.chdir('/Users/kkfluous/Downloads')
+
+SRC = '租赁任务考核_2026年5月.xlsx'
+BAK = '租赁任务考核_2026年5月_原版备份.xlsx'
+
+# ─── Styles ───
+DARK_BLUE  = PatternFill('solid', fgColor='1F4E78')
+LIGHT_BLUE = PatternFill('solid', fgColor='D9E1F2')
+HD_BLUE    = PatternFill('solid', fgColor='4472C4')
+YELLOW_BG  = PatternFill('solid', fgColor='FFF2CC')
+ALT_GRAY   = PatternFill('solid', fgColor='F2F2F2')
+YELLOW_HL  = PatternFill('solid', fgColor='FFFF00')
+NO_FILL    = PatternFill('solid', fgColor='00000000')
+
+TITLE_F   = Font(bold=True, size=14, color='FFFFFF')
+SEC_F     = Font(bold=True, size=12, color='1F4E78')
+SUB_F     = Font(bold=True, size=11, color='1F4E78')
+HDR_F     = Font(bold=True, size=11, color='FFFFFF')
+BODY_F    = Font(size=11)
+BOLD_F    = Font(bold=True, size=11)
+RED_F     = Font(size=10, color='FF0000')
+DESC_F    = Font(size=10)
+
+THIN = Side('thin', color='B4B4B4')
+BORD = Border(THIN, THIN, THIN, THIN)
+CC   = Alignment(horizontal='center', vertical='center', wrap_text=True)
+LL   = Alignment(horizontal='left',   vertical='center', wrap_text=True)
+LT   = Alignment(horizontal='left',   vertical='top',    wrap_text=True)
+
+# ─── Load yellow intervention plates ───
+wb_i = openpyxl.load_workbook('标黄为5月干预车辆.xlsx')
+ws_i = wb_i['Sheet1']
+YELLOW_PLATES = set()
+for row in ws_i.iter_rows(min_row=2, values_only=False):
+    cell = row[0]
+    fill = cell.fill
+    is_y = (fill and fill.patternType == 'solid' and fill.start_color
+            and fill.start_color.rgb and 'FFFF00' in str(fill.start_color.rgb))
+    if is_y:
+        p = str(cell.value).strip() if cell.value else ''
+        if p: YELLOW_PLATES.add(p)
+wb_i.close()
+# 手动移除4辆已确认非干预
+YELLOW_PLATES.difference_update({'粤A03186F', '粤A01128F', '粤A08391F', '粤A01396F'})
+print(f"Yellow plates: {len(YELLOW_PLATES)} (removed 4)")
+
+# ─── Load data ───
+def load_recs(fp):
+    wb = openpyxl.load_workbook(fp, data_only=True)
+    ws = wb['业务考核视图']
+    h = [c.value for c in next(ws.iter_rows(min_row=1, max_row=1))]
+    recs = []
+    for row in ws.iter_rows(min_row=2, values_only=True):
+        if row[0]: recs.append(dict(zip(h, row)))
+    wb.close()
+    return recs, h
+
+src_recs, src_h = load_recs(SRC)
+bak_recs, bak_h = load_recs(BAK)
+
+print(f"Source: {len(src_recs)} rows, cols={len(src_h)}")
+print(f"Backup: {len(bak_recs)} rows, cols={len(bak_h)}")
+
+# Identify intervention records by plate (yellow list)
+intv_recs = []
+for sr in src_recs:
+    plate = str(sr['车牌号']).strip()
+    if plate not in YELLOW_PLATES:
+        continue
+
+    gps   = float(sr.get('实际行驶里程(km)', 0) or 0)
+    tbox  = float(sr.get('原TBOX里程(km)', 0) or 0)
+    test  = float(sr.get('测试里程(km)', 0) or 0)
+    tgt   = float(sr.get('应考核里程(km)', 0) or 0)
+
+    # Find backup counterpart
+    key = (plate, str(sr.get('销售经理','')), str(sr.get('项目名称','')))
+    br = None
+    for b in bak_recs:
+        bk = (str(b['车牌号']).strip(), str(b.get('销售经理','')), str(b.get('项目名称','')))
+        if bk == key: br = b; break
+    old_actual = float(br['实际行驶里程(km)'] or 0) if br else 0
+    old_pass   = str(br['是否达标']).strip() if br else '?'
+    new_pass   = str(sr.get('是否达标','')).strip()
+    flip = ''
+    if old_pass == '达标' and new_pass == '未达标': flip = '达→未达标'
+    elif old_pass == '未达标' and new_pass == '达标': flip = '未→达标'
+    intv_recs.append(dict(plate=plate, dept=sr.get('部门名称',''), sales=sr.get('销售经理',''),
+                          client=sr.get('客户名称',''), target_name=sr.get('考核目标',''),
+                          tgt_km=tgt, tbox=tbox, gps=gps, test=test, flip=flip,
+                          days=sr.get('考核天数'), rate=sr.get('完成率(%)'),
+                          new_pass=new_pass, old_pass=old_pass))
+
+print(f"Intervention records: {len(intv_recs)}")
+flip_down = [r for r in intv_recs if r['flip'] == '达→未达标']
+flip_up   = [r for r in intv_recs if r['flip'] == '未→达标']
+print(f"达→未达标: {len(flip_down)}, 未→达标: {len(flip_up)}")
+
+# ─── Reference column layout (O=GPS, P=TBOX, Q=Test) ───
+# Build full records with columns in reference order
+REF_HEADERS = [
+    '车牌号', '考核年份', '考核月份', '部门名称', '销售经理', '客户名称',
+    '合同编号', '项目名称', '考核目标ID', '考核目标', '考核开始日期',
+    '考核结束日期', '考核天数', '应考核里程(km)',
+    '实际行驶里程(km)',     # O = GPS (考核口径)
+    '原TBOX里程(km)',       # P = TBOX (仪表盘)
+    '测试里程(km)',         # Q = test
+    '完成率(%)', '是否达标', '考核状态',
+    '交车日期', '预计还车日期', '实际还车日期',
+]
+
+def reorder_rec(src_rec):
+    """Map source record to reference column order. O=GPS, P=TBOX, Q=Test.
+    For non-intv: O=仪表盘, P=空, Q=0. For intv: O=GPS, P=仪表盘, Q=GPS-仪表盘."""
+    out = {}
+    plate = str(src_rec.get('车牌号','')).strip()
+    is_intv = plate in YELLOW_PLATES
+    for h in REF_HEADERS:
+        if h == '实际行驶里程(km)':
+            # O column: always use source's actual (which is仪表盘 for non-intv, GPS for intv)
+            out[h] = src_rec.get('实际行驶里程(km)')
+        elif h == '原TBOX里程(km)':
+            # P column: only fill if intv, else empty
+            out[h] = src_rec.get('原TBOX里程(km)') if is_intv else None
+        elif h == '测试里程(km)':
+            out[h] = src_rec.get('测试里程(km)') if is_intv else 0
+        else:
+            out[h] = src_rec.get(h)
+    return out
+
+all_ref = [reorder_rec(r) for r in src_recs]
+intv_ref = [reorder_rec(r) for r in src_recs if str(r['车牌号']).strip() in YELLOW_PLATES]
+
+# ─── Build workbook ───
+wb = openpyxl.Workbook()
+
+def hdr_row(ws, row, ncols):
+    for c in range(1, ncols+1):
+        cell = ws.cell(row=row, column=c)
+        cell.fill = HD_BLUE; cell.font = HDR_F; cell.alignment = CC; cell.border = BORD
+
+def dat_row(ws, row, ncols, alt=False):
+    for c in range(1, ncols+1):
+        cell = ws.cell(row=row, column=c)
+        cell.font = BODY_F; cell.alignment = CC; cell.border = BORD
+        if alt: cell.fill = ALT_GRAY
+
+# ═══════════════════════════════
+# Sheets 2,3,4 first (so Sheet1 formulas can reference them)
+# ═══════════════════════════════
+
+def write_data_sheet(wb, name, records, highlight_intv=False):
+    ws = wb.create_sheet(name)
+    for ci, h in enumerate(REF_HEADERS, 1):
+        ws.cell(row=1, column=ci, value=h)
+    hdr_row(ws, 1, len(REF_HEADERS))
+    for ri, rec in enumerate(records):
+        r = ri + 2
+        is_intv = str(rec.get('车牌号','')).strip() in YELLOW_PLATES
+        for ci, h in enumerate(REF_HEADERS, 1):
+            val = rec.get(h)
+            if val is not None:
+                ws.cell(row=r, column=ci, value=val)
+            ws.cell(row=r, column=ci).border = BORD
+            ws.cell(row=r, column=ci).font = Font(size=10)
+            ws.cell(row=r, column=ci).alignment = CC
+        if highlight_intv and is_intv:
+            for ci in range(1, len(REF_HEADERS)+1):
+                ws.cell(row=r, column=ci).fill = YELLOW_HL
+    for ci in range(1, len(REF_HEADERS)+1):
+        ws.column_dimensions[get_column_letter(ci)].width = 12
+    return ws
+
+# Sheet 2: 业务考核视图 (all records, intv highlighted yellow)
+ws2 = write_data_sheet(wb, '业务考核视图', all_ref, highlight_intv=True)
+
+# Sheet 3: 5月业务考核车辆总览 (all records, no highlight)
+ws3 = write_data_sheet(wb, '5月业务考核车辆总览', all_ref, highlight_intv=False)
+
+# Sheet 4: 涉及测试的业务考核车辆 (intv only)
+ws4 = write_data_sheet(wb, '涉及测试的业务考核车辆', intv_ref, highlight_intv=False)
+
+# ═══════════════════════════════
+# Sheet 1: 测试与考核情况说明 (with formulas)
+# ═══════════════════════════════
+ws1 = wb.active
+ws1.title = '测试与考核情况说明'
+
+MAX_R = 500  # generous row range for formulas
+
+# R1: Title
+ws1.merge_cells('A1:G1')
+c = ws1.cell(row=1, column=1, value='2026年5月 业务考核 — 测试与考核情况说明')
+c.fill = DARK_BLUE; c.font = TITLE_F; c.alignment = CC
+ws1.row_dimensions[1].height = 32
+
+# R2-R3: Description
+ws1.merge_cells('A2:G3')
+desc = ('5月业务考核车辆中，部分车辆当月存在测试行为（仪表盘里程 > GPS 实际行驶里程），其差额作为测试里程剥离，不参与考核。\n'
+        '对于有测试行为的车辆，使用GPS里程作为实际业务考核里程并参与绩效计算。\n'
+        '对于无5月GPS数据的干预车辆，实际行驶里程记为0，判定未达标。\n'
+        '下表数据由公式实时计算，源sheet数据更新后自动刷新。')
+c = ws1.cell(row=2, column=1, value=desc)
+c.fill = YELLOW_BG; c.font = DESC_F; c.alignment = LT
+ws1.row_dimensions[2].height = 72
+
+# R4: empty spacer
+# R5: Section 一
+ws1.merge_cells('A5:G5')
+c = ws1.cell(row=5, column=1, value='一、5月业务考核与干预情况')
+c.fill = LIGHT_BLUE; c.font = SEC_F; c.alignment = LL
+ws1.row_dimensions[5].height = 24
+
+# R6: Table header
+for ci, h in enumerate(['项目', '5月业务考核车辆', '涉及测试的车辆', '占比', '', '', ''], 1):
+    ws1.cell(row=6, column=ci, value=h)
+hdr_row(ws1, 6, 4)
+ws1.row_dimensions[6].height = 24
+
+# R7-R11: Formula rows
+# Sheet references for formulas (openpyxl writes them as-is)
+S3 = "'5月业务考核车辆总览'"
+S4 = "'涉及测试的业务考核车辆'"
+
+formula_rows = [
+    # (row, label, B_formula, C_formula, B_fmt, C_fmt, D_fmt)
+    (7, '车辆数(辆)',
+     f'=SUMPRODUCT(({S3}!A2:A{MAX_R}<>"")/(COUNTIF({S3}!A2:A{MAX_R},{S3}!A2:A{MAX_R})+({S3}!A2:A{MAX_R}="")))',
+     f'=SUMPRODUCT(({S4}!A2:A{MAX_R}<>"")/(COUNTIF({S4}!A2:A{MAX_R},{S4}!A2:A{MAX_R})+({S4}!A2:A{MAX_R}="")))',
+     '#,##0', '#,##0', '0.00%'),
+    (8, '产生考核记录数(条)',
+     f'=COUNTA({S3}!A2:A{MAX_R})',
+     f'=COUNTA({S4}!A2:A{MAX_R})',
+     '#,##0', '#,##0', '0.00%'),
+    (9, '仪表盘里程(km)',
+     f'=SUM({S3}!O2:O{MAX_R})+SUM({S4}!Q2:Q{MAX_R})',
+     f'=SUM({S4}!O2:O{MAX_R})+SUM({S4}!Q2:Q{MAX_R})',
+     '#,##0.0', '#,##0.0', '0.00%'),
+    (10, '业务考核里程 GPS(km)',
+     f'=SUM({S3}!O2:O{MAX_R})',
+     f'=SUM({S4}!O2:O{MAX_R})',
+     '#,##0.0', '#,##0.0', '0.00%'),
+    (11, '测试里程(km)',
+     f'=SUM({S4}!Q2:Q{MAX_R})',
+     f'=SUM({S4}!Q2:Q{MAX_R})',
+     '#,##0.0', '#,##0.0', '0.00%'),
+]
+
+for ri, (row_num, label, b_f, c_f, b_fmt, c_fmt, d_fmt) in enumerate(formula_rows):
+    r = row_num
+    ws1.cell(row=r, column=1, value=label).font = BOLD_F
+    ws1.cell(row=r, column=1).alignment = LL
+    ws1.cell(row=r, column=1).border = BORD
+
+    ws1.cell(row=r, column=2, value=b_f).number_format = b_fmt
+    ws1.cell(row=r, column=2).font = BODY_F
+    ws1.cell(row=r, column=2).alignment = CC
+    ws1.cell(row=r, column=2).border = BORD
+
+    ws1.cell(row=r, column=3, value=c_f).number_format = c_fmt
+    ws1.cell(row=r, column=3).font = BODY_F
+    ws1.cell(row=r, column=3).alignment = CC
+    ws1.cell(row=r, column=3).border = BORD
+
+    ws1.cell(row=r, column=4, value=f'=IFERROR(C{r}/B{r},0)').number_format = d_fmt
+    ws1.cell(row=r, column=4).font = BODY_F
+    ws1.cell(row=r, column=4).alignment = CC
+    ws1.cell(row=r, column=4).border = BORD
+
+    if ri % 2 == 1:
+        for c in range(1, 5):
+            ws1.cell(row=r, column=c).fill = ALT_GRAY
+    ws1.row_dimensions[r].height = 22
+
+# R12: empty spacer
+# R13: Section 二
+ws1.merge_cells('A13:G13')
+c = ws1.cell(row=13, column=1, value='二、干预车辆考核达标情况说明')
+c.fill = LIGHT_BLUE; c.font = SEC_F; c.alignment = LL
+ws1.row_dimensions[13].height = 24
+
+# R14: Summary description
+ws1.merge_cells('A14:G14')
+flip_desc = (f'共 {len(intv_recs)} 条考核记录涉及里程干预；替换为GPS里程后，'
+             f'共 {len(flip_down)} 条考核车辆记录由"达标"转判为"未达标"。按车型分类如下：')
+c = ws1.cell(row=14, column=1, value=flip_desc)
+c.font = RED_F; c.alignment = LL
+ws1.row_dimensions[14].height = 28
+
+# Group by vehicle type
+by_target = defaultdict(lambda: defaultdict(lambda: defaultdict(int)))
+flip_by_target = defaultdict(lambda: defaultdict(lambda: defaultdict(int)))
+for r in intv_recs:
+    by_target[r['target_name']][r['dept']][r['sales']] += 1
+for r in flip_down:
+    flip_by_target[r['target_name']][r['dept']][r['sales']] += 1
+
+target_order = sorted(by_target.keys(), key=lambda t: -sum(sum(s.values()) for s in flip_by_target.get(t, {}).values()))
+
+row = 15
+for tgt in target_order:
+    dept_sales = flip_by_target.get(tgt, {})
+    total_count = sum(sum(s.values()) for s in by_target[tgt].values())
+    flip_count = sum(sum(s.values()) for s in dept_sales.values())
+
+    if flip_count == 0: continue  # only show types with flips
+
+    # Vehicle type title
+    if flip_count > 0:
+        title = f'▎{tgt}（{flip_count} 辆）'
+    else:
+        title = f'▎{tgt}（{total_count} 辆）'
+    ws1.merge_cells(start_row=row, start_column=1, end_row=row, end_column=7)
+    c = ws1.cell(row=row, column=1, value=title)
+    c.fill = LIGHT_BLUE; c.font = SUB_F; c.alignment = LL
+    for ci in range(1, 8): ws1.cell(row=row, column=ci).border = BORD
+    ws1.row_dimensions[row].height = 22
+    row += 1
+
+    # Headers (two columns side-by-side: A-C and E-G)
+    for base in [1, 5]:
+        for ci, h in enumerate(['部门', '销售经理', '数量'], base):
+            ws1.cell(row=row, column=ci, value=h)
+    for ci in range(1, 8):
+        ws1.cell(row=row, column=ci).fill = HD_BLUE
+        ws1.cell(row=row, column=ci).font = HDR_F
+        ws1.cell(row=row, column=ci).alignment = CC
+        ws1.cell(row=row, column=ci).border = BORD
+    ws1.row_dimensions[row].height = 22
+    row += 1
+
+    # Flatten sales list sorted by department then count
+    dept_order = sorted(dept_sales.keys(), key=lambda d: -sum(dept_sales[d].values()))
+    sales_list = []
+    for dept in dept_order:
+        for sales in sorted(dept_sales[dept].keys()):
+            sales_list.append((dept, sales, dept_sales[dept][sales]))
+
+    n = len(sales_list)
+    left  = sales_list[:(n+1)//2]
+    right = sales_list[(n+1)//2:]
+
+    for ri2 in range(max(len(left), len(right))):
+        if ri2 < len(left):
+            d, s, cnt = left[ri2]
+            ws1.cell(row=row, column=1, value=d)
+            ws1.cell(row=row, column=2, value=s)
+            ws1.cell(row=row, column=3, value=cnt)
+        if ri2 < len(right):
+            d, s, cnt = right[ri2]
+            ws1.cell(row=row, column=5, value=d)
+            ws1.cell(row=row, column=6, value=s)
+            ws1.cell(row=row, column=7, value=cnt)
+        for ci in range(1, 8):
+            cell = ws1.cell(row=row, column=ci)
+            cell.border = BORD; cell.font = BODY_F; cell.alignment = CC
+            if ri2 % 2 == 1: cell.fill = ALT_GRAY
+        ws1.row_dimensions[row].height = 22
+        row += 1
+
+    row += 1  # blank between vehicle types
+
+# Section 三
+ws1.merge_cells(start_row=row, start_column=1, end_row=row, end_column=7)
+c = ws1.cell(row=row, column=1, value='三、里程三列数据来源说明')
+c.fill = LIGHT_BLUE; c.font = SEC_F; c.alignment = LL
+ws1.row_dimensions[row].height = 24
+row += 1
+
+for ci, h in enumerate(['列', '字段名', '数据来源', '说明', '', '', ''], 1):
+    ws1.cell(row=row, column=ci, value=h)
+hdr_row(ws1, row, 4)
+ws1.row_dimensions[row].height = 24
+row += 1
+
+ref_data = [
+    ('O', '实际行驶里程(km)', 'GPS 轨迹还原',
+     '基于 G7 GPS 定位轨迹按考核期逐日汇总计算的运营里程（total_mileage/100000=km），本月里程考核口径。干预车辆已替换为GPS里程，非干预车辆沿用原仪表盘里程。'),
+    ('P', '原TBOX里程(km)', '车辆 T-Box / 仪表盘读数',
+     '当月仪表总里程（业务系统原始数据），含运营+测试+调试里程。仅干预车辆填充，非干预车辆为空。'),
+    ('Q', '测试里程(km)', 'P − O（原TBOX − GPS，差额）',
+     '仪表盘跑了但 GPS 未记录到的部分，判定为厂内测试/调试/路试等非运营里程，不计入考核。仅干预车辆填充，非干预车辆为0。'),
+]
+
+for ri, (col_letter, name, source, note) in enumerate(ref_data):
+    ws1.cell(row=row, column=1, value=col_letter).font = BOLD_F
+    ws1.cell(row=row, column=1).alignment = CC
+    ws1.cell(row=row, column=2, value=name)
+    ws1.cell(row=row, column=3, value=source)
+    ws1.cell(row=row, column=4, value=note).alignment = LL
+    dat_row(ws1, row, 4, alt=(ri%2==1))
+    ws1.cell(row=row, column=1).font = BOLD_F
+    ws1.row_dimensions[row].height = 42
+    row += 1
+
+# Column widths
+for ci, w in enumerate([12, 22, 13, 12, 13, 22, 13], 1):
+    ws1.column_dimensions[get_column_letter(ci)].width = w
+
+# ─── Save ───
+out = 'new-kpi-202605_1/租赁任务考核_2026年5月_干预情况说明_Final.xlsx'
+wb.save(out)
+print(f'✅ {out} ({len(wb.sheetnames)} sheets)')
diff --git a/_gen_resignation_may.py b/_gen_resignation_may.py
new file mode 100644
index 0000000..467a143
--- /dev/null
+++ b/_gen_resignation_may.py
@@ -0,0 +1,375 @@
+#!/usr/bin/env python3
+"""重新生成 离职人员5月奖金处理明细.xlsx — 当月转嫁，其他不发了"""
+import os, openpyxl, warnings
+from openpyxl.styles import Font, PatternFill, Alignment, Border, Side
+from openpyxl.utils import get_column_letter
+from collections import defaultdict
+from calc_engine import *
+
+warnings.filterwarnings('ignore')
+os.chdir('/Users/kkfluous/Downloads')
+
+# ── Full calc ──
+print("读取源数据...")
+D = {}
+for m in [1,2,3,4,5]:
+    D[m] = read_file(f'租赁任务考核_2026年{m}月.xlsx', m)
+G = {m: grp(D[m], m) for m in [1,2,3,4,5]}
+calc_jan_carryover(G[1])
+feb_data = calc_feb(G[1], G[2])
+mar_data = calc_mar(G[1], G[2], G[3], feb_data)
+apr_data = calc_apr(G[1], G[2], G[3], G[4], mar_data)
+may_data = calc_may(G[1], G[2], G[3], G[4], G[5], apr_data)
+
+RESIGNATIONS = {
+    '赵连飞': {'dept': '业务二部', 'transfer_to': '刘念念'},
+    '伍仲文': {'dept': '业务六部', 'transfer_to': '钟祥'},
+    '岑彦':   {'dept': '业务六部', 'transfer_to': '钟祥'},
+}
+resigned = set(RESIGNATIONS.keys())
+
+# ── Collect resigned payments ──
+detail_records = []
+for ptype, items in may_data.items():
+    for it in items:
+        if it['销售'] not in resigned:
+            continue
+        name = it['销售']
+        is_dangyue = (ptype == '当月')
+        transfer_to = RESIGNATIONS[name]['transfer_to']
+        handling = f'转{transfer_to}(需确认)' if is_dangyue else '不发了'
+        transfer_disp = transfer_to if is_dangyue else '—'
+        note = '5月当月达标，转新销售' if is_dangyue else f'{ptype}，不发了'
+
+        # Get additional info from G5
+        g5 = G[5].get((it['车牌'], name))
+        client = ''
+        for m in [1,2,3,4,5]:
+            for r in D[m]:
+                if str(r['车牌号']).strip() == it['车牌'] and r['销售经理'] == name:
+                    client = r.get('客户名称', '')
+                    break
+            if client: break
+
+        monthly_bonus = g5['奖励额'] if g5 else 0
+        days = g5['天数'] if g5 else ''
+        target_name = ''
+        if g5 and g5.get('recs'):
+            target_name = g5['recs'][0].get('考核目标', '')
+
+        # 盈亏查询月
+        if ptype == '当月' or ptype == '累计补发5月':
+            loss_month = 5
+        elif ptype == '结转':
+            loss_month = 5
+        elif ptype.startswith('补发'):
+            loss_month = int(ptype.replace('补发','').replace('月',''))
+        else:
+            loss_month = 5
+
+        detail_records.append({
+            'name': name, 'plate': it['车牌'], 'pay_type': ptype,
+            'amount': it['额'], 'handling': handling, 'transfer': transfer_disp,
+            'note': note, 'client': client, 'loss_month': loss_month,
+            'loss_status': '否', 'monthly_bonus': monthly_bonus,
+            'days': days, 'target': target_name,
+        })
+
+print(f"离职人员发放记录: {len(detail_records)} 条")
+
+# ── Styles ──
+HEAD_FILL = PatternFill('solid', fgColor='4472C4')
+HEAD_FONT = Font(bold=True, color='FFFFFF', size=10)
+SUB_FILL  = PatternFill('solid', fgColor='D9E1F2')
+ALT_FILL  = PatternFill('solid', fgColor='F2F2F2')
+TITLE_FNT = Font(bold=True, size=14)
+NOTE_FNT  = Font(size=10, color='C00000')
+NORM_FNT  = Font(size=10)
+BOLD_FNT  = Font(bold=True, size=10)
+THIN = Side('thin', color='B4B4B4')
+BORD = Border(THIN, THIN, THIN, THIN)
+CC = Alignment(horizontal='center', vertical='center', wrap_text=True)
+LL = Alignment(horizontal='left', vertical='center', wrap_text=True)
+
+def hdr(ws, r, nc):
+    for c in range(1, nc+1):
+        cell = ws.cell(row=r, column=c)
+        cell.fill = HEAD_FILL; cell.font = HEAD_FONT; cell.alignment = CC; cell.border = BORD
+
+def drw(ws, r, nc, alt=False):
+    for c in range(1, nc+1):
+        cell = ws.cell(row=r, column=c)
+        cell.font = NORM_FNT; cell.alignment = CC; cell.border = BORD
+        if alt: cell.fill = ALT_FILL
+
+wb = openpyxl.Workbook()
+
+# ═══════════════════════════════════════
+# Sheet 1: 离职人员发放处理总览
+# ═══════════════════════════════════════
+ws1 = wb.active
+ws1.title = '离职人员发放处理总览'
+
+ws1.merge_cells('A1:G1')
+ws1.cell(row=1, column=1, value='2026年5月核算 — 离职人员奖金处理方案').font = TITLE_FNT
+ws1.cell(row=1, column=1).alignment = CC
+ws1.row_dimensions[1].height = 30
+
+ws1.merge_cells('A2:G2')
+ws1.cell(row=2, column=1,
+    value='处理规则：仅5月当月达标奖金转给新接手的销售（需与其他部门确认是否发放）；其余结转、补发、累计补发等一律不发了。').font = NOTE_FNT
+ws1.cell(row=2, column=1).alignment = LL
+ws1.row_dimensions[2].height = 22
+
+r = 4
+oh = ['离职人员', '部门', '业务转嫁至', '不发了', '转新销售(需确认)', '原考核应发合计']
+for c, h in enumerate(oh, 1):
+    ws1.cell(row=r, column=c, value=h)
+hdr(ws1, r, 6)
+r += 1
+
+total_not = 0; total_trf = 0
+for name in ['赵连飞', '伍仲文', '岑彦']:
+    info = RESIGNATIONS[name]
+    recs = [p for p in detail_records if p['name'] == name]
+    not_pay = sum(p['amount'] for p in recs if p['pay_type'] != '当月')
+    transfer = sum(p['amount'] for p in recs if p['pay_type'] == '当月')
+    total_not += not_pay; total_trf += transfer
+    ws1.cell(row=r, column=1, value=name)
+    ws1.cell(row=r, column=2, value=info['dept'])
+    ws1.cell(row=r, column=3, value=info['transfer_to'])
+    ws1.cell(row=r, column=4, value=round(not_pay, 2))
+    ws1.cell(row=r, column=5, value=round(transfer, 2))
+    ws1.cell(row=r, column=6, value=round(not_pay+transfer, 2))
+    drw(ws1, r, 6)
+    ws1.row_dimensions[r].height = 22
+    r += 1
+
+# 合计
+for c in range(1, 7):
+    ws1.cell(row=r, column=c).fill = SUB_FILL
+    ws1.cell(row=r, column=c).border = BORD
+    ws1.cell(row=r, column=c).font = BOLD_FNT
+    ws1.cell(row=r, column=c).alignment = CC
+ws1.cell(row=r, column=1, value='合计')
+ws1.cell(row=r, column=4, value=round(total_not, 2))
+ws1.cell(row=r, column=5, value=round(total_trf, 2))
+ws1.cell(row=r, column=6, value=round(total_not+total_trf, 2))
+
+r += 2
+ws1.merge_cells(start_row=r, start_column=1, end_row=r, end_column=6)
+ws1.cell(row=r, column=1, value='转嫁接收方汇总（需与其他部门确认是否发放）').font = Font(bold=True, size=12, color='1F4E78')
+ws1.cell(row=r, column=1).alignment = CC
+r += 1
+
+for c, hh in enumerate(['接收人', '来源离职人员', '转嫁金额', '备注'], 1):
+    ws1.cell(row=r, column=c, value=hh)
+hdr(ws1, r, 4)
+r += 1
+
+by_recv = defaultdict(lambda: {'sources':[], 'amt':0})
+for rec in detail_records:
+    if rec['pay_type'] == '当月':
+        t = rec['transfer']
+        by_recv[t]['sources'].append(rec['name'])
+        by_recv[t]['amt'] += rec['amount']
+
+for receiver in ['刘念念', '钟祥']:
+    d = by_recv[receiver]
+    ws1.cell(row=r, column=1, value=receiver)
+    ws1.cell(row=r, column=2, value='、'.join(sorted(set(d['sources']))))
+    ws1.cell(row=r, column=3, value=round(d['amt'], 2))
+    ws1.cell(row=r, column=4, value='需与相关部门确认是否发放给新销售')
+    drw(ws1, r, 4)
+    r += 1
+
+for i, w in enumerate([14, 22, 18, 18, 18, 18], 1):
+    ws1.column_dimensions[get_column_letter(i)].width = w
+
+# ═══════════════════════════════════════
+# Sheet 2: 发放明细(逐条)
+# ═══════════════════════════════════════
+ws2 = wb.create_sheet('发放明细(逐条)')
+ws2.merge_cells('A1:M1')
+ws2.cell(row=1, column=1, value='离职人员 — 5月考核应发逐条明细').font = TITLE_FNT
+ws2.cell(row=1, column=1).alignment = CC
+ws2.row_dimensions[1].height = 28
+
+dh = ['离职人员', '车牌号', '发放类型', '考核应发', '盈亏查询月', '客户名称',
+      '客户盈亏', '处理方式', '转嫁接收人', '5月考核天数', '月度奖励', '考核目标', '备注']
+r2 = 3
+for c, h in enumerate(dh, 1):
+    ws2.cell(row=r2, column=c, value=h)
+hdr(ws2, r2, 13)
+r2 += 1
+
+name_order = ['赵连飞', '伍仲文', '岑彦']
+detail_records.sort(key=lambda x: (name_order.index(x['name']), 0 if x['pay_type']=='当月' else 1, x['plate']))
+
+for i, rec in enumerate(detail_records):
+    ws2.cell(row=r2, column=1, value=rec['name'])
+    ws2.cell(row=r2, column=2, value=rec['plate'])
+    ws2.cell(row=r2, column=3, value=rec['pay_type'])
+    ws2.cell(row=r2, column=4, value=round(rec['amount'], 2))
+    ws2.cell(row=r2, column=5, value=f"{rec['loss_month']}月")
+    ws2.cell(row=r2, column=6, value=rec['client'])
+    ws2.cell(row=r2, column=7, value=rec['loss_status'])
+    ws2.cell(row=r2, column=8, value=rec['handling'])
+    ws2.cell(row=r2, column=9, value=rec['transfer'])
+    ws2.cell(row=r2, column=10, value=rec['days'] if rec['days'] else '')
+    ws2.cell(row=r2, column=11, value=rec['monthly_bonus'] if rec['monthly_bonus'] else '')
+    ws2.cell(row=r2, column=12, value=rec['target'])
+    ws2.cell(row=r2, column=13, value=rec['note'])
+    drw(ws2, r2, 13, alt=(i%2==1))
+    ws2.row_dimensions[r2].height = 22
+    r2 += 1
+
+# 合计行
+ws2.cell(row=r2, column=3, value='合计')
+ws2.cell(row=r2, column=4, value=round(sum(r['amount'] for r in detail_records), 2))
+for c in range(1, 14):
+    ws2.cell(row=r2, column=c).fill = SUB_FILL
+    ws2.cell(row=r2, column=c).border = BORD
+    ws2.cell(row=r2, column=c).font = BOLD_FNT
+    ws2.cell(row=r2, column=c).alignment = CC
+
+for i, w in enumerate([10, 12, 14, 14, 10, 24, 8, 18, 14, 12, 10, 22, 28], 1):
+    ws2.column_dimensions[get_column_letter(i)].width = w
+
+# ═══════════════════════════════════════
+# Sheets 3-5: 个人明细
+# ═══════════════════════════════════════
+for name, title_txt, transfer_to in [
+    ('赵连飞', '赵连飞（2026-04-30离职，业务转→刘念念）', '刘念念'),
+    ('伍仲文', '伍仲文（2026-05-09离职，业务转→钟祥）', '钟祥'),
+    ('岑彦', '岑彦（2026-05-09离职，业务转→钟祥）', '钟祥'),
+]:
+    ws = wb.create_sheet(name)
+    ws.merge_cells('A1:K1')
+    ws.cell(row=1, column=1, value=title_txt).font = TITLE_FNT
+    ws.cell(row=1, column=1).alignment = CC
+    ws.row_dimensions[1].height = 28
+
+    ph = ['车牌号', '发放类型', '考核应发', '盈亏查询月', '客户名称', '客户盈亏',
+          '处理方式', '5月考核天数', '月度奖励', '考核目标', '备注']
+    r3 = 3
+    for c, h in enumerate(ph, 1):
+        ws.cell(row=r3, column=c, value=h)
+    hdr(ws, r3, 11)
+    r3 += 1
+
+    person_recs = [p for p in detail_records if p['name'] == name]
+    dangyue_recs = [p for p in person_recs if p['pay_type'] == '当月']
+    other_recs = [p for p in person_recs if p['pay_type'] != '当月']
+
+    for i, rec in enumerate(person_recs):
+        h_disp = rec['handling']
+        ws.cell(row=r3, column=1, value=rec['plate'])
+        ws.cell(row=r3, column=2, value=rec['pay_type'])
+        ws.cell(row=r3, column=3, value=round(rec['amount'], 2))
+        ws.cell(row=r3, column=4, value=f"{rec['loss_month']}月")
+        ws.cell(row=r3, column=5, value=rec['client'])
+        ws.cell(row=r3, column=6, value=rec['loss_status'])
+        ws.cell(row=r3, column=7, value=h_disp)
+        ws.cell(row=r3, column=8, value=rec['days'] if rec['days'] else '')
+        ws.cell(row=r3, column=9, value=rec['monthly_bonus'] if rec['monthly_bonus'] else '')
+        ws.cell(row=r3, column=10, value=rec['target'])
+        ws.cell(row=r3, column=11, value=rec['note'])
+        drw(ws, r3, 11, alt=(i%2==1))
+        ws.row_dimensions[r3].height = 22
+        r3 += 1
+
+    dn = sum(p['amount'] for p in dangyue_recs)
+    ot = sum(p['amount'] for p in other_recs)
+    gt = dn + ot
+
+    # 不发了小计
+    ws.cell(row=r3, column=2, value='不发了小计').font = BOLD_FNT
+    ws.cell(row=r3, column=3, value=round(ot, 2)).font = BOLD_FNT
+    for c in range(1, 12):
+        ws.cell(row=r3, column=c).fill = SUB_FILL; ws.cell(row=r3, column=c).border = BORD
+    ws.cell(row=r3, column=2).alignment = CC; ws.cell(row=r3, column=3).alignment = CC
+    r3 += 1
+
+    # 转嫁小计
+    ws.cell(row=r3, column=2, value=f'转{transfer_to}小计').font = BOLD_FNT
+    ws.cell(row=r3, column=3, value=round(dn, 2)).font = BOLD_FNT
+    ws.cell(row=r3, column=11, value='需与其他部门确认').font = Font(size=9, color='C00000')
+    for c in range(1, 12):
+        ws.cell(row=r3, column=c).fill = SUB_FILL; ws.cell(row=r3, column=c).border = BORD
+    ws.cell(row=r3, column=2).alignment = CC; ws.cell(row=r3, column=3).alignment = CC
+    r3 += 1
+
+    # 合计
+    ws.cell(row=r3, column=2, value='合计').font = BOLD_FNT
+    ws.cell(row=r3, column=3, value=round(gt, 2)).font = BOLD_FNT
+    for c in range(1, 12):
+        ws.cell(row=r3, column=c).fill = SUB_FILL; ws.cell(row=r3, column=c).border = BORD
+    ws.cell(row=r3, column=2).alignment = CC; ws.cell(row=r3, column=3).alignment = CC
+
+    for i, w in enumerate([12, 14, 14, 10, 26, 8, 18, 12, 10, 22, 30], 1):
+        ws.column_dimensions[get_column_letter(i)].width = w
+
+# ═══════════════════════════════════════
+# Sheets 6-7: 转嫁明细
+# ═══════════════════════════════════════
+for title_txt, sources, receiver in [
+    ('转嫁至刘念念的奖金 — 来源：赵连飞(2026-04-30离职)', ['赵连飞'], '刘念念'),
+    ('转嫁至钟祥的奖金 — 来源：伍仲文(2026-05-09离职)、岑彦(2026-05-09离职)', ['伍仲文', '岑彦'], '钟祥'),
+]:
+    ws = wb.create_sheet(f'转{receiver}明细')
+    ws.merge_cells('A1:K1')
+    ws.cell(row=1, column=1, value=title_txt).font = TITLE_FNT
+    ws.cell(row=1, column=1).alignment = CC
+    ws.row_dimensions[1].height = 28
+
+    ws.merge_cells('A2:K2')
+    ws.cell(row=2, column=1, value='⚠ 以下金额需与其他部门确认是否可发放给新销售').font = NOTE_FNT
+    ws.cell(row=2, column=1).alignment = LL
+
+    th = ['来源离职人员', '车牌号', '发放类型', '考核应发', '盈亏查询月', '客户名称',
+          '客户盈亏', '5月考核天数', '月度奖励', '考核目标', '备注']
+    r4 = 4
+    for c, h in enumerate(th, 1):
+        ws.cell(row=r4, column=c, value=h)
+    hdr(ws, r4, 11)
+    r4 += 1
+
+    trf_recs = [p for p in detail_records if p['name'] in sources and p['pay_type'] == '当月']
+    for i, rec in enumerate(trf_recs):
+        ws.cell(row=r4, column=1, value=rec['name'])
+        ws.cell(row=r4, column=2, value=rec['plate'])
+        ws.cell(row=r4, column=3, value='当月')
+        ws.cell(row=r4, column=4, value=round(rec['amount'], 2))
+        ws.cell(row=r4, column=5, value=f"{rec['loss_month']}月")
+        ws.cell(row=r4, column=6, value=rec['client'])
+        ws.cell(row=r4, column=7, value=rec['loss_status'])
+        ws.cell(row=r4, column=8, value=rec['days'] if rec['days'] else '')
+        ws.cell(row=r4, column=9, value=rec['monthly_bonus'] if rec['monthly_bonus'] else '')
+        ws.cell(row=r4, column=10, value=rec['target'])
+        ws.cell(row=r4, column=11, value='需确认是否发给新销售')
+        drw(ws, r4, 11, alt=(i%2==1))
+        ws.row_dimensions[r4].height = 22
+        r4 += 1
+
+    ws.cell(row=r4, column=3, value='合计')
+    ws.cell(row=r4, column=4, value=round(sum(r['amount'] for r in trf_recs), 2))
+    for c in range(1, 12):
+        ws.cell(row=r4, column=c).fill = SUB_FILL; ws.cell(row=r4, column=c).border = BORD
+        ws.cell(row=r4, column=c).font = BOLD_FNT; ws.cell(row=r4, column=c).alignment = CC
+
+    for i, w in enumerate([14, 12, 14, 14, 10, 26, 8, 12, 10, 22, 22], 1):
+        ws.column_dimensions[get_column_letter(i)].width = w
+
+# ── Save ──
+out = 'new-kpi-202605_1/离职人员5月奖金处理明细.xlsx'
+wb.save(out)
+
+print(f'\n✅ {out} ({len(wb.sheetnames)} sheets)')
+print(f'\n汇总:')
+for name in ['赵连飞', '伍仲文', '岑彦']:
+    recs = [p for p in detail_records if p['name'] == name]
+    not_pay = sum(p['amount'] for p in recs if p['pay_type'] != '当月')
+    trf = sum(p['amount'] for p in recs if p['pay_type'] == '当月')
+    print(f'  {name}: 不发了={not_pay:.2f}  转{RESIGNATIONS[name]["transfer_to"]}={trf:.2f}  合计={not_pay+trf:.2f}')
+print(f'  总计: 不发了={total_not:.2f}  转新销售={total_trf:.2f}')
diff --git a/_process_may_intervention.py b/_process_may_intervention.py
new file mode 100644
index 0000000..65b5964
--- /dev/null
+++ b/_process_may_intervention.py
@@ -0,0 +1,275 @@
+#!/usr/bin/env python3
+"""Process May intervention: replace mileage with GPS, add TBOX/test columns."""
+import os, shutil, openpyxl, pymysql, warnings
+from openpyxl.styles import Font, PatternFill, Alignment, Border, Side
+from openpyxl.utils import get_column_letter
+from datetime import datetime, date, timedelta
+from collections import defaultdict
+
+warnings.filterwarnings('ignore')
+os.chdir('/Users/kkfluous/Downloads')
+
+SRC = '租赁任务考核_2026年5月.xlsx'
+BAK = '租赁任务考核_2026年5月_原版备份.xlsx'
+OUT = SRC  # overwrite
+
+# ── Step 1: Backup ──
+if not os.path.exists(BAK):
+    shutil.copy2(SRC, BAK)
+    print(f"✅ Backup: {BAK}")
+else:
+    print(f"跳过备份（已存在）")
+
+# ── Step 2: Load intervention plates ──
+wb_i = openpyxl.load_workbook('标黄为5月干预车辆.xlsx')
+ws_i = wb_i['Sheet1']
+yellow_plates = set()
+for row in ws_i.iter_rows(min_row=2, values_only=False):
+    cell = row[0]
+    fill = cell.fill
+    is_yellow = (fill and fill.patternType == 'solid' and fill.start_color
+                 and fill.start_color.rgb and 'FFFF00' in str(fill.start_color.rgb))
+    if is_yellow:
+        p = str(cell.value).strip() if cell.value else ''
+        if p:
+            yellow_plates.add(p)
+wb_i.close()
+print(f"干预车辆: {len(yellow_plates)}")
+
+# ── Step 3: Read source records ──
+wb_s = openpyxl.load_workbook(SRC, data_only=True)
+ws_s = wb_s['业务考核视图']
+src_h = [c.value for c in next(ws_s.iter_rows(min_row=1, max_row=1))]
+rows_raw = list(ws_s.iter_rows(min_row=2, values_only=True))
+wb_s.close()
+
+records = []
+for row in rows_raw:
+    if not row[0]:
+        continue
+    records.append(dict(zip(src_h, row)))
+print(f"源数据: {len(records)} 条")
+
+# ── Step 4: GPS from DB ──
+print("查询数据库...")
+conn = pymysql.connect(host='101.133.130.65', user='root', password='ln123!@#',
+                       database='hydrogen_energy', charset='utf8mb4')
+cur = conn.cursor()
+cur.execute("SELECT plate_number, dates, total_mileage FROM ln_vehicle_g7_mileage WHERE dates >= '2026-05-01' AND dates <= '2026-06-08'")
+db_rows = cur.fetchall()
+conn.close()
+
+# Group: plate -> {date: mileage}
+pmap = defaultdict(lambda: defaultdict(int))
+for p, d, m in db_rows:
+    dd = d if isinstance(d, date) else (d.date() if isinstance(d, datetime) else date.fromisoformat(str(d)[:10]))
+    pmap[p][dd] = m or 0
+
+print(f"DB plates loaded: {len(pmap)}")
+
+# Compute GPS km per plate per考核期
+def get_gps_km(plate, start_v, end_v):
+    """GPS里程(km) = sum of daily total_mileage / 100000"""
+    if plate not in pmap:
+        return 0.0
+
+    # Parse dates
+    if isinstance(start_v, datetime):
+        sd = start_v.date()
+    elif isinstance(start_v, date):
+        sd = start_v
+    else:
+        return 0.0
+
+    if isinstance(end_v, datetime):
+        ed = end_v.date()
+    elif isinstance(end_v, date):
+        ed = end_v
+    else:
+        return 0.0
+
+    total = 0
+    d = sd
+    daily = pmap[plate]
+    while d <= ed:
+        total += daily.get(d, 0)
+        d += timedelta(days=1)
+    return total / 100000.0
+
+# Build plate -> gps for intervention vehicles
+plate_gps = {}
+for rec in records:
+    p = str(rec['车牌号']).strip()
+    if p not in yellow_plates:
+        continue
+    if p in plate_gps:
+        continue
+    km = get_gps_km(p, rec.get('考核开始日期'), rec.get('考核结束日期'))
+    plate_gps[p] = round(km, 2)
+
+print(f"GPS数据获取: {len(plate_gps)}/{len(yellow_plates)}")
+gps_zero = sum(1 for v in plate_gps.values() if v == 0)
+print(f"GPS=0: {gps_zero} 辆")
+
+# ── Step 5: Generate new workbook ──
+YELLOW = PatternFill('solid', fgColor='FFFF00')
+RED = PatternFill('solid', fgColor='FF0000')
+REDF = Font(color='FFFFFF', bold=True)
+HFILL = PatternFill('solid', fgColor='4472C4')
+HFONT = Font(bold=True, color='FFFFFF', size=10)
+BORD = Border(Side('thin','B4B4B4'), Side('thin','B4B4B4'), Side('thin','B4B4B4'), Side('thin','B4B4B4'))
+CC = Alignment(horizontal='center', vertical='center', wrap_text=True)
+NFONT = Font(size=10)
+
+wb = openpyxl.Workbook()
+ws = wb.active
+ws.title = '业务考核视图'
+
+# New header layout:
+# 车牌号(0) 考核年份(1) 考核月份(2) 部门名称(3) 销售经理(4) 客户名称(5) 合同编号(6) 项目名称(7)
+# 考核目标ID(8) 考核目标(9) 考核开始日期(10) 考核结束日期(11) 考核天数(12) 应考核里程(13)
+# 原TBOX里程(14) 实际行驶里程=GPS(15) 测试里程(16) 完成率(17) 是否达标(18) 考核状态(19) ...
+# Column map: src_col -> dst_col (0-indexed)
+# 0-13 stay same, 14(old actual) -> split into 14(TBOX) 15(new actual) 16(test), 15(old rate) -> 17, 16(old pass)->18, 17(status)->19, 18-20 -> 20-22
+
+new_h = ['车牌号', '考核年份', '考核月份', '部门名称', '销售经理', '客户名称',
+         '合同编号', '项目名称', '考核目标ID', '考核目标', '考核开始日期',
+         '考核结束日期', '考核天数', '应考核里程(km)',
+         '原TBOX里程(km)', '实际行驶里程(km)', '测试里程(km)',
+         '完成率(%)', '是否达标', '考核状态',
+         '交车日期', '预计还车日期', '实际还车日期']
+
+for ci, h in enumerate(new_h):
+    c = ci + 1
+    cell = ws.cell(row=1, column=c, value=h)
+    cell.fill = HFILL
+    cell.font = HFONT
+    cell.alignment = CC
+    cell.border = BORD
+
+# Write data
+for ri, rec in enumerate(records):
+    r = ri + 2
+    plate = str(rec['车牌号']).strip()
+    is_intv = plate in yellow_plates
+
+    # Columns 0-13: copy directly (0-indexed in src_h)
+    for si in range(14):  # 车牌号 ~ 应考核里程(km)
+        val = rec.get(src_h[si])
+        if val is not None:
+            ws.cell(row=r, column=si + 1, value=val)
+        ws.cell(row=r, column=si + 1).border = BORD
+        ws.cell(row=r, column=si + 1).alignment = CC
+        ws.cell(row=r, column=si + 1).font = NFONT
+
+    # Column 14 (0-indexed): 原TBOX里程(km) - old actual mileage
+    tbox = float(rec.get('实际行驶里程(km)', 0) or 0)
+    ws.cell(row=r, column=15, value=tbox).border = BORD
+    ws.cell(row=r, column=15).alignment = CC
+    ws.cell(row=r, column=15).font = NFONT
+
+    # Column 15 (0-indexed:16): 实际行驶里程(km) - GPS for intv, TBOX for others
+    target_km = float(rec.get('应考核里程(km)', 0) or 0)
+    if is_intv:
+        new_actual = plate_gps.get(plate, 0)
+    else:
+        new_actual = tbox
+
+    cell_actual = ws.cell(row=r, column=16, value=new_actual)
+    cell_actual.border = BORD
+    cell_actual.alignment = CC
+    cell_actual.font = NFONT
+
+    # Column 16 (0-indexed:17): 测试里程(km) = TBOX - GPS
+    test_km = round(tbox - new_actual, 2) if is_intv else 0
+    ws.cell(row=r, column=17, value=test_km).border = BORD
+    ws.cell(row=r, column=17).alignment = CC
+    ws.cell(row=r, column=17).font = NFONT
+
+    # Column 17 (0-indexed:18): 完成率(%)
+    rate = round(new_actual / target_km * 100, 2) if target_km > 0 else 0
+    ws.cell(row=r, column=18, value=rate).border = BORD
+    ws.cell(row=r, column=18).alignment = CC
+    ws.cell(row=r, column=18).font = NFONT
+
+    # Column 18 (0-indexed:19): 是否达标
+    new_pass = '达标' if new_actual >= target_km and target_km > 0 else '未达标'
+    ws.cell(row=r, column=19, value=new_pass).border = BORD
+    ws.cell(row=r, column=19).alignment = CC
+    ws.cell(row=r, column=19).font = NFONT
+
+    # Column 19 (0-indexed:20): 考核状态
+    val = rec.get('考核状态')
+    ws.cell(row=r, column=20, value=val if val else '进行中').border = BORD
+    ws.cell(row=r, column=20).alignment = CC
+    ws.cell(row=r, column=20).font = NFONT
+
+    # Columns 20-22 (0-indexed): 交车日期, 预计还车日期, 实际还车日期
+    for si in range(18, 21):
+        val = rec.get(src_h[si])
+        if val is not None:
+            ws.cell(row=r, column=si + 1 + 5, value=val)  # +5 offset for new cols
+        else:
+            ws.cell(row=r, column=si + 1 + 5, value='')
+        ws.cell(row=r, column=si + 1 + 5).border = BORD
+        ws.cell(row=r, column=si + 1 + 5).alignment = CC
+        ws.cell(row=r, column=si + 1 + 5).font = NFONT
+
+    # ── Formatting: yellow for intervention, red for GPS=0 ──
+    if is_intv:
+        for c in range(1, len(new_h) + 1):
+            cell = ws.cell(row=r, column=c)
+            # Don't override red cells
+            existing_fill = cell.fill
+            is_red = (existing_fill.patternType == 'solid' and
+                      existing_fill.start_color and
+                      existing_fill.start_color.rgb and
+                      'FF0000' in str(existing_fill.start_color.rgb))
+            if not is_red:
+                cell.fill = YELLOW
+            else:
+                # Red stays, but also yellow background? No, red means GPS=0, keep it
+                pass
+
+        # Red highlight for GPS=0
+        if plate_gps.get(plate, 0) == 0:
+            ws.cell(row=r, column=16).fill = RED  # 实际行驶里程 column
+            ws.cell(row=r, column=16).font = REDF
+
+# Column widths
+widths = [12, 8, 8, 12, 10, 26, 18, 22, 10, 22, 13, 13, 10, 14, 16, 16, 14, 12, 10, 8, 12, 12, 12]
+for i, w in enumerate(widths):
+    ws.column_dimensions[get_column_letter(i+1)].width = w
+
+# ── Save ──
+wb.save(OUT)
+print(f"\n✅ 已保存: {OUT} ({len(records)} rows, {len(new_h)} cols)")
+
+# ── Stats ──
+old_p = sum(1 for r in records if str(r.get('是否达标','')).strip() == '达标')
+new_p = 0
+flip_down = []
+flip_up = []
+for rec in records:
+    p = str(rec['车牌号']).strip()
+    tgt = float(rec.get('应考核里程(km)', 0) or 0)
+    old_status = str(rec.get('是否达标','')).strip()
+    if p in yellow_plates:
+        actual = plate_gps.get(p, 0)
+    else:
+        actual = float(rec.get('实际行驶里程(km)', 0) or 0)
+    n = '达标' if actual >= tgt and tgt > 0 else '未达标'
+    if n == '达标':
+        new_p += 1
+    if old_status == '达标' and n == '未达标':
+        flip_down.append((p, float(rec.get('实际行驶里程(km)',0) or 0), actual, tgt))
+    elif old_status == '未达标' and n == '达标':
+        flip_up.append((p, float(rec.get('实际行驶里程(km)',0) or 0), actual, tgt))
+
+print(f"达标变化: {old_p} -> {new_p} ({new_p - old_p:+d})")
+print(f"达→未达标: {len(flip_down)} 辆")
+for p, tb, gp, tg in flip_down[:10]:
+    print(f"  {p} TBOX={tb:.1f} GPS={gp:.2f} 目标={tg:.0f}")
+print(f"未→达标: {len(flip_up)} 辆")
+for p, tb, gp, tg in flip_up[:10]:
+    print(f"  {p} TBOX={tb:.1f} GPS={gp:.2f} 目标={tg:.0f}")
diff --git a/add_vehicle_type_to_july_summary.py b/add_vehicle_type_to_july_summary.py
new file mode 100644
index 0000000..280e6df
--- /dev/null
+++ b/add_vehicle_type_to_july_summary.py
@@ -0,0 +1,70 @@
+from copy import copy
+from pathlib import Path
+import shutil
+
+from openpyxl import load_workbook
+
+
+OUTPUT = Path('/Users/kkfluous/Downloads/7月核算结果/7月考核车辆多源里程汇总.xlsx')
+BACKUP = Path('/Users/kkfluous/Downloads/7月核算结果/7月考核车辆多源里程汇总_增加车型前.xlsx')
+SOURCE = Path('/Users/kkfluous/Downloads/7月核算结果/租赁任务考核_2026年7月.xlsx')
+
+
+def plate(value):
+    return ''.join(str(value or '').upper().split())
+
+
+src_wb = load_workbook(SOURCE, read_only=True, data_only=True)
+src_ws = src_wb['业务考核视图']
+headers = [c.value for c in src_ws[1]]
+idx = {name: i for i, name in enumerate(headers)}
+types = {}
+for row in src_ws.iter_rows(min_row=2, values_only=True):
+    if not row[idx['车牌号']]:
+        continue
+    start = row[idx['考核开始日期']]
+    end = row[idx['考核结束日期']]
+    key = (plate(row[idx['车牌号']]), start.date(), end.date(), str(row[idx['客户名称']] or '').strip())
+    types[key] = row[idx['考核目标']] or '无来源'
+src_wb.close()
+
+if not BACKUP.exists():
+    shutil.copy2(OUTPUT, BACKUP)
+
+wb = load_workbook(OUTPUT)
+ws = wb.active
+header_row = next(r for r in range(1, 8) if ws.cell(r, 2).value == '车牌')
+data_start = header_row + 1
+
+for merged in list(ws.merged_cells.ranges):
+    if str(merged) in ('A1:L1', 'A1:M1'):
+        ws.unmerge_cells(str(merged))
+ws.insert_cols(4, 1)
+ws.cell(header_row, 4).value = '车型'
+ws.cell(header_row, 4)._style = copy(ws.cell(header_row, 3)._style)
+ws.cell(header_row, 4).font = copy(ws.cell(header_row, 3).font)
+ws.cell(header_row, 4).fill = copy(ws.cell(header_row, 3).fill)
+ws.cell(header_row, 4).border = copy(ws.cell(header_row, 3).border)
+ws.cell(header_row, 4).alignment = copy(ws.cell(header_row, 3).alignment)
+
+missing = []
+for row in range(data_start, ws.max_row + 1):
+    start = ws.cell(row, 5).value
+    end = ws.cell(row, 6).value
+    key = (plate(ws.cell(row, 2).value), start.date(), end.date(), str(ws.cell(row, 3).value or '').strip())
+    ws.cell(row, 4).value = types.get(key, '无来源')
+    ws.cell(row, 4)._style = copy(ws.cell(row, 3)._style)
+    ws.cell(row, 4).font = copy(ws.cell(row, 3).font)
+    ws.cell(row, 4).fill = copy(ws.cell(row, 3).fill)
+    ws.cell(row, 4).border = copy(ws.cell(row, 3).border)
+    ws.cell(row, 4).alignment = copy(ws.cell(row, 3).alignment)
+    if key not in types:
+        missing.append(key)
+
+ws.merge_cells('A1:M1')
+ws.column_dimensions['D'].width = 24
+ws.auto_filter.ref = f'A{header_row}:M{ws.max_row}'
+ws.print_title_rows = f'1:{header_row}'
+wb.save(OUTPUT)
+
+print({'records': ws.max_row - header_row, 'type_missing': len(missing), 'gps_column': ws.cell(header_row, 13).value})
diff --git a/audit_july_daily_test_rule.py b/audit_july_daily_test_rule.py
new file mode 100644
index 0000000..e079919
--- /dev/null
+++ b/audit_july_daily_test_rule.py
@@ -0,0 +1,90 @@
+from ast import parse, walk, Call, Constant, keyword
+from collections import defaultdict
+from datetime import datetime, date
+from pathlib import Path
+from zipfile import ZipFile
+from xml.etree import ElementTree as ET
+import re
+
+import openpyxl
+import pymysql
+
+BASE = Path('/Users/kkfluous/Downloads')
+ASSESSMENT = BASE / '7月核算结果/租赁任务考核_2026年7月.xlsx'
+TBOX = BASE / '车辆里程查询_20260701-20260731_1024辆.xlsx'
+DETAILS = [BASE / '里程统计[天][2026-07-01至2026-07-31].xlsx', BASE / '里程统计[天][2026-07-01至2026-07-31] (1).xlsx']
+NS = '{http://schemas.openxmlformats.org/spreadsheetml/2006/main}'
+
+def norm(v): return re.sub(r'\s+', '', str(v or '')).upper()
+
+def cfg():
+    tree=parse((BASE/'replace_gps_jun.py').read_text())
+    for node in walk(tree):
+        if isinstance(node,Call) and getattr(node.func,'attr',None)=='connect':
+            d={x.arg:x.value.value for x in node.keywords if isinstance(x.value,Constant)}
+            if {'host','user','password','database'}<=d.keys(): return d
+    raise RuntimeError('no db config')
+
+def read_details():
+    out=defaultdict(dict)
+    for path in DETAILS:
+        with ZipFile(path) as z:
+            root=ET.fromstring(z.read('xl/worksheets/sheet1.xml'))
+            for row in root.findall(f'.//{NS}row'):
+                if int(row.get('r'))<4: continue
+                d={}
+                for c in row.findall(f'{NS}c'):
+                    col=re.match(r'[A-Z]+',c.get('r')).group()
+                    if c.get('t')=='inlineStr': d[col]=''.join(t.text or '' for t in c.findall(f'.//{NS}t'))
+                    else:
+                        v=c.find(f'{NS}v'); d[col]=v.text if v is not None else None
+                p=norm(d.get('H')); ds=str(d.get('O') or '').strip()
+                if p and p!='合计' and re.fullmatch(r'2026-07-\d{2}',ds):
+                    out[p][datetime.strptime(ds,'%Y-%m-%d').date()]=float(d.get('P') or 0)
+    return out
+
+awb=openpyxl.load_workbook(ASSESSMENT,data_only=True)
+aws=awb['业务考核视图']
+h=[c.value for c in aws[1]]; ix={v:i for i,v in enumerate(h)}
+candidates=set(); records=[]
+for ri,row in enumerate(aws.iter_rows(min_row=2,values_only=True),2):
+    fills={aws.cell(ri,c).fill.fgColor.rgb for c in range(1,aws.max_column+1) if aws.cell(ri,c).fill.fill_type=='solid'}
+    is_candidate=any(str(x).upper().endswith('FFFF00') for x in fills if x)
+    p=norm(row[ix['车牌号']]); start=row[ix['考核开始日期']].date(); end=row[ix['考核结束日期']].date()
+    records.append((p,start,end,is_candidate))
+    if is_candidate:candidates.add(p)
+awb.close()
+
+twb=openpyxl.load_workbook(TBOX,read_only=True,data_only=True); tws=twb['里程查询']
+daycols={i:v.date() for i,v in enumerate(next(tws.iter_rows(min_row=6,max_row=6,values_only=True))) if isinstance(v,datetime)}
+tbox=defaultdict(dict)
+for row in tws.iter_rows(min_row=7,values_only=True):
+    p=norm(row[0])
+    if p:
+        for i,d in daycols.items():
+            if row[i] is not None:tbox[p][d]=float(row[i])
+twb.close()
+
+gps=read_details()
+dc=cfg();dc.update(charset='utf8mb4',connect_timeout=8,read_timeout=30,write_timeout=30)
+conn=pymysql.connect(**dc)
+try:
+    with conn.cursor() as cur:
+        ph=','.join(['%s']*len(candidates))
+        cur.execute(f'SELECT plate_number,dates,total_mileage FROM ln_vehicle_g7_mileage WHERE dates >= %s AND dates <= %s AND plate_number IN ({ph})',[date(2026,7,1),date(2026,7,31),*sorted(candidates)])
+        rows=cur.fetchall()
+finally:conn.close()
+for p,d,m in rows:
+    if isinstance(d,datetime):d=d.date()
+    gps[norm(p)].setdefault(d,float(m or 0)/100000)
+
+candidate_days=0; comparable=0; flagged=0; plates_no_tbox=set(); plates_no_gps=set(); flagged_plates=set()
+for p,start,end,cand in records:
+    if not cand:continue
+    for n in range((end-start).days+1):
+        d=start.fromordinal(start.toordinal()+n);candidate_days+=1
+        if d not in tbox.get(p,{}):plates_no_tbox.add(p);continue
+        if d not in gps.get(p,{}):plates_no_gps.add(p);continue
+        comparable+=1
+        if tbox[p][d]-gps[p][d]>50:flagged+=1;flagged_plates.add(p)
+print({'candidate_records':sum(x[3] for x in records),'candidate_plates':len(candidates),'candidate_days':candidate_days,'comparable_days':comparable,'flagged_days':flagged,'flagged_plates':len(flagged_plates),'plates_with_any_tbox_gap':len(plates_no_tbox),'plates_with_any_gps_gap':len(plates_no_gps),'no_tbox_sample':sorted(plates_no_tbox)[:20],'no_gps_sample':sorted(plates_no_gps)[:20]})
diff --git a/audit_july_gps_database.py b/audit_july_gps_database.py
new file mode 100644
index 0000000..1097ebd
--- /dev/null
+++ b/audit_july_gps_database.py
@@ -0,0 +1,59 @@
+from ast import parse, walk, Call, Constant, keyword
+from datetime import date
+from pathlib import Path
+import re
+
+import openpyxl
+import pymysql
+
+
+def connection_kwargs_from_existing_script():
+    tree = parse(Path('/Users/kkfluous/Downloads/replace_gps_jun.py').read_text())
+    for node in walk(tree):
+        if not isinstance(node, Call):
+            continue
+        func = node.func
+        if getattr(func, 'attr', None) != 'connect':
+            continue
+        values = {}
+        for item in node.keywords:
+            if isinstance(item, keyword) and isinstance(item.value, Constant):
+                values[item.arg] = item.value.value
+        if {'host', 'user', 'password', 'database'} <= values.keys():
+            return values
+    raise RuntimeError('未在既有脚本中找到数据库连接配置')
+
+
+src = openpyxl.load_workbook(
+    '/Users/kkfluous/Downloads/7月核算结果/租赁任务考核_2026年7月.xlsx',
+    read_only=True,
+    data_only=True,
+)['业务考核视图']
+headers = [c.value for c in src[1]]
+plate_col = headers.index('车牌号')
+plates = sorted({re.sub(r'\s+', '', str(r[plate_col] or '')).upper() for r in src.iter_rows(min_row=2, values_only=True) if r[plate_col]})
+
+cfg = connection_kwargs_from_existing_script()
+cfg.update(charset='utf8mb4', connect_timeout=8, read_timeout=30, write_timeout=30)
+conn = pymysql.connect(**cfg)
+try:
+    with conn.cursor() as cur:
+        placeholders = ','.join(['%s'] * len(plates))
+        cur.execute(
+            f'''SELECT plate_number, dates, total_mileage
+                FROM ln_vehicle_g7_mileage
+                WHERE dates >= %s AND dates <= %s
+                  AND plate_number IN ({placeholders})''',
+            [date(2026, 7, 1), date(2026, 7, 31), *plates],
+        )
+        rows = cur.fetchall()
+finally:
+    conn.close()
+
+db_plates = {re.sub(r'\s+', '', str(r[0] or '')).upper() for r in rows}
+print({
+    'assessment_plates': len(plates),
+    'database_rows': len(rows),
+    'database_plates': len(db_plates),
+    'uncovered_plates': len(set(plates) - db_plates),
+})
diff --git a/audit_recalculate_july_from_multi_source.py b/audit_recalculate_july_from_multi_source.py
new file mode 100644
index 0000000..f1a1775
--- /dev/null
+++ b/audit_recalculate_july_from_multi_source.py
@@ -0,0 +1,49 @@
+import openpyxl
+import re
+from collections import Counter
+from datetime import datetime
+from pathlib import Path
+
+NEW = Path('/Users/kkfluous/Downloads/7月考核车辆多源里程汇总_用户确认版.xlsx')
+OLD = Path('/Users/kkfluous/Downloads/7月核算结果/租赁任务考核_2026年7月.xlsx')
+
+def norm(v):
+    return re.sub(r'\s+', '', str(v or '')).upper()
+
+def day(v):
+    return v.date() if isinstance(v, datetime) else v
+
+nw = openpyxl.load_workbook(NEW, data_only=True, read_only=False)
+ns = nw['7月考核车辆里程汇总']
+nh = {ns.cell(2, c).value: c for c in range(1, ns.max_column + 1)}
+ow = openpyxl.load_workbook(OLD, data_only=True, read_only=False)
+os = ow['业务考核视图']
+oh = {os.cell(1, c).value: c for c in range(1, os.max_column + 1)}
+
+N = [(norm(ns.cell(r, nh['车牌']).value), day(ns.cell(r, nh['考核开始日期']).value),
+      day(ns.cell(r, nh['考核结束日期']).value), ns.cell(r, nh['核算里程(km)']).value)
+     for r in range(3, ns.max_row + 1)]
+O = [(norm(os.cell(r, oh['车牌号']).value), day(os.cell(r, oh['考核开始日期']).value),
+      day(os.cell(r, oh['考核结束日期']).value), os.cell(r, oh['实际行驶里程(km)']).value)
+     for r in range(2, os.max_row + 1)]
+nk, ok = Counter(x[:3] for x in N), Counter(x[:3] for x in O)
+nd = {x[:3]: x[3] for x in N}
+changes = []
+for x in O:
+    nv, ov = nd.get(x[:3]), x[3]
+    if isinstance(nv, (int, float)) and abs(float(nv) - float(ov or 0)) > 0.005:
+        changes.append((*x[:3], ov, nv, nv - float(ov or 0)))
+
+print({
+    'new_rows': len(N), 'old_rows': len(O),
+    'new_dup_keys': sum(v > 1 for v in nk.values()), 'old_dup_keys': sum(v > 1 for v in ok.values()),
+    'missing_in_new': len(set(ok) - set(nk)), 'extra_in_new': len(set(nk) - set(ok)),
+    'changed_rows': len(changes),
+    'old_total': round(sum(float(x[3] or 0) for x in O), 2),
+    'new_total': round(sum(float(x[3] or 0) for x in N if isinstance(x[3], (int, float))), 2),
+})
+print('missing', list(set(ok) - set(nk))[:10])
+print('extra', list(set(nk) - set(ok))[:10])
+for x in sorted(changes, key=lambda x: abs(x[5]), reverse=True)[:15]:
+    print(x)
+
diff --git a/backfill_july_all_gps.py b/backfill_july_all_gps.py
new file mode 100644
index 0000000..c2ba204
--- /dev/null
+++ b/backfill_july_all_gps.py
@@ -0,0 +1,101 @@
+from ast import parse, walk, Call, Constant, keyword
+from collections import defaultdict
+from datetime import datetime, date
+from pathlib import Path
+import re
+import shutil
+
+import openpyxl
+import pymysql
+
+
+OUTPUT = Path('/Users/kkfluous/Downloads/7月核算结果/7月考核车辆多源里程汇总.xlsx')
+BACKUP = Path('/Users/kkfluous/Downloads/7月核算结果/7月考核车辆多源里程汇总_GPS全量回补前.xlsx')
+
+
+def norm(value):
+    return re.sub(r'\s+', '', str(value or '')).upper()
+
+
+def db_config():
+    tree = parse(Path('/Users/kkfluous/Downloads/replace_gps_jun.py').read_text())
+    for node in walk(tree):
+        if not isinstance(node, Call) or getattr(node.func, 'attr', None) != 'connect':
+            continue
+        values = {}
+        for item in node.keywords:
+            if isinstance(item, keyword) and isinstance(item.value, Constant):
+                values[item.arg] = item.value.value
+        if {'host', 'user', 'password', 'database'} <= values.keys():
+            return values
+    raise RuntimeError('未找到既有GPS数据库配置')
+
+
+wb = openpyxl.load_workbook(OUTPUT)
+ws = wb.active
+header_row = next(r for r in range(1, 8) if ws.cell(r, 12).value == 'GPS里程(km)')
+data_start = header_row + 1
+plates = sorted({norm(ws.cell(r, 2).value) for r in range(data_start, ws.max_row + 1) if ws.cell(r, 2).value})
+
+cfg = db_config()
+cfg.update(charset='utf8mb4', connect_timeout=8, read_timeout=30, write_timeout=30)
+conn = pymysql.connect(**cfg)
+try:
+    with conn.cursor() as cur:
+        placeholders = ','.join(['%s'] * len(plates))
+        cur.execute(
+            f'''SELECT plate_number, dates, total_mileage
+                FROM ln_vehicle_g7_mileage
+                WHERE dates >= %s AND dates <= %s
+                  AND plate_number IN ({placeholders})''',
+            [date(2026, 7, 1), date(2026, 7, 31), *plates],
+        )
+        db_rows = cur.fetchall()
+finally:
+    conn.close()
+
+daily = defaultdict(dict)
+for plate, day, raw_mileage in db_rows:
+    if isinstance(day, datetime):
+        day = day.date()
+    daily[norm(plate)][day] = float(raw_mileage or 0) / 100000.0
+
+if not BACKUP.exists():
+    shutil.copy2(OUTPUT, BACKUP)
+
+filled = []
+still_missing = []
+for row in range(data_start, ws.max_row + 1):
+    cell = ws.cell(row, 12)
+    if cell.value != '无来源':
+        continue
+    plate = norm(ws.cell(row, 2).value)
+    start = ws.cell(row, 4).value
+    end = ws.cell(row, 5).value
+    if isinstance(start, datetime):
+        start = start.date()
+    if isinstance(end, datetime):
+        end = end.date()
+    matched = [value for day, value in daily.get(plate, {}).items() if start <= day <= end]
+    if matched:
+        cell.value = round(sum(matched), 2)
+        cell.number_format = '#,##0.0'
+        filled.append((plate, start, end, cell.value))
+    else:
+        still_missing.append((plate, start, end))
+
+ws['A1'] = (
+    '1. 纯氢里程(km)、纯电里程(km)来源于广州平台统计，TBOX里程来源于车机上报的仪表盘里程统计，'
+    'GPS里程优先使用两份7月GPS日明细，缺失车辆由GPS数据库每日里程按考核区间回补；测试车辆以7月最终考核源表已确认GPS为准\n'
+    '2. 广州平台总里程(km)=纯氢里程(km)+纯电里程(km)\n'
+    '3. 已确认广州平台暂未接入86辆普货车数据；所有GPS来源均无对应记录时标记为“无来源”'
+)
+wb.save(OUTPUT)
+
+print({
+    'database_rows': len(db_rows),
+    'database_plates': len(daily),
+    'backfilled_records': len(filled),
+    'remaining_missing_records': len(still_missing),
+    'remaining_missing_plates': len({x[0] for x in still_missing}),
+})
diff --git a/build_july_multi_source_summary.py b/build_july_multi_source_summary.py
new file mode 100644
index 0000000..4482f93
--- /dev/null
+++ b/build_july_multi_source_summary.py
@@ -0,0 +1,216 @@
+from collections import defaultdict
+from datetime import datetime
+from pathlib import Path
+from zipfile import ZipFile
+from xml.etree import ElementTree as ET
+import re
+
+from openpyxl import Workbook, load_workbook
+from openpyxl.styles import Alignment, Border, Font, PatternFill, Side
+from openpyxl.utils import get_column_letter
+
+
+BASE = Path("/Users/kkfluous/Downloads")
+ASSESSMENT = BASE / "7月核算结果/租赁任务考核_2026年7月.xlsx"
+PLATFORM_DIR = BASE / "广州新能源平台_纯氢纯电里程_2026年7月_按日"
+GPS_FILES = [
+    BASE / "里程统计[天][2026-07-01至2026-07-31].xlsx",
+    BASE / "里程统计[天][2026-07-01至2026-07-31] (1).xlsx",
+]
+OUTPUT = BASE / "7月核算结果/7月考核车辆多源里程汇总.xlsx"
+NS = "{http://schemas.openxmlformats.org/spreadsheetml/2006/main}"
+
+
+def normalize_plate(value):
+    return re.sub(r"\s+", "", str(value or "")).upper()
+
+
+def parse_gps_file(path):
+    result = {}
+    with ZipFile(path) as zf:
+        root = ET.fromstring(zf.read("xl/worksheets/sheet1.xml"))
+        for row in root.findall(f".//{NS}row"):
+            if int(row.get("r")) < 4:
+                continue
+            values = {}
+            for cell in row.findall(f"{NS}c"):
+                col = re.match(r"[A-Z]+", cell.get("r")).group()
+                if cell.get("t") == "inlineStr":
+                    values[col] = "".join(t.text or "" for t in cell.findall(f".//{NS}t"))
+                else:
+                    node = cell.find(f"{NS}v")
+                    values[col] = node.text if node is not None else None
+            plate = normalize_plate(values.get("H"))
+            date_text = str(values.get("O") or "").strip()
+            if not plate or plate == "合计" or not re.fullmatch(r"2026-07-\d{2}", date_text):
+                continue
+            result[(plate, datetime.strptime(date_text, "%Y-%m-%d").date())] = float(values.get("P") or 0)
+    return result
+
+
+platform = defaultdict(lambda: [0.0, 0.0])
+platform_presence = set()
+for path in sorted(PLATFORM_DIR.glob("2026-07-??_纯氢纯电里程.xlsx")):
+    wb = load_workbook(path, read_only=True, data_only=True)
+    ws = wb.active
+    for row in ws.iter_rows(min_row=2, values_only=True):
+        plate = normalize_plate(row[5])
+        date_match = re.search(r"2026-07-\d{2}", str(row[6] or ""))
+        if not plate or not date_match:
+            continue
+        date = datetime.strptime(date_match.group(), "%Y-%m-%d").date()
+        electric = float(row[7] or 0)
+        hydrogen = float(row[8] or 0)
+        platform[(plate, date)][0] += hydrogen
+        platform[(plate, date)][1] += electric
+        platform_presence.add((plate, date))
+    wb.close()
+
+gps = {}
+for path in GPS_FILES:
+    incoming = parse_gps_file(path)
+    overlap = set(gps) & set(incoming)
+    if overlap:
+        raise RuntimeError(f"GPS来源存在重复车辆日期：{sorted(overlap)[:5]}")
+    gps.update(incoming)
+
+source_wb = load_workbook(ASSESSMENT, read_only=True, data_only=True)
+source_ws = source_wb["业务考核视图"]
+headers = [c.value for c in source_ws[1]]
+idx = {name: i for i, name in enumerate(headers)}
+
+records = []
+for row in source_ws.iter_rows(min_row=2, values_only=True):
+    plate = normalize_plate(row[idx["车牌号"]])
+    if not plate:
+        continue
+    start = row[idx["考核开始日期"]].date()
+    end = row[idx["考核结束日期"]].date()
+    dates = [(start.fromordinal(start.toordinal() + i)) for i in range((end - start).days + 1)]
+
+    p_dates = [d for d in dates if (plate, d) in platform_presence]
+    if p_dates:
+        hydrogen = sum(platform[(plate, d)][0] for d in p_dates)
+        electric = sum(platform[(plate, d)][1] for d in p_dates)
+    else:
+        hydrogen = electric = None
+
+    test_mileage = row[idx["测试里程(km)"]]
+    is_test = isinstance(test_mileage, (int, float)) and test_mileage > 0
+    g_dates = [d for d in dates if (plate, d) in gps]
+    gps_total = sum(gps[(plate, d)] for d in g_dates) if g_dates else None
+    # 7月最终考核源表的“实际行驶里程”已按原处理链路写入测试车辆GPS里程：
+    # 优先智能管车日报，缺失时用两份GPS日明细补充。补充明细未命中时，
+    # 必须从最终源表回填，不能误标为“无来源”。
+    if gps_total is None and is_test:
+        gps_total = row[idx["实际行驶里程(km)"]]
+    records.append({
+        "plate": plate,
+        "customer": row[idx["客户名称"]] or "",
+        "start": start,
+        "end": end,
+        "days": (end - start).days + 1,
+        "test": "是" if is_test else "否",
+        "hydrogen": hydrogen,
+        "electric": electric,
+        "tbox": row[idx["原TBOX里程(km)"]] if row[idx["原TBOX里程(km)"]] is not None else None,
+        "gps": gps_total,
+    })
+source_wb.close()
+
+wb = Workbook()
+ws = wb.active
+ws.title = "7月考核车辆里程汇总"
+columns = [
+    "序号", "车牌", "客户", "考核开始日期", "考核结束日期", "考核天数", "是否存在测试",
+    "纯氢里程(km)", "纯电里程(km)", "广州平台总里程(km)", "TBOX里程(km)", "GPS里程(km)",
+]
+ws.append(columns)
+
+for n, rec in enumerate(records, 1):
+    row_num = ws.max_row + 1
+    ws.append([
+        n, rec["plate"], rec["customer"], rec["start"], rec["end"], rec["days"], rec["test"],
+        rec["hydrogen"] if rec["hydrogen"] is not None else "无来源",
+        rec["electric"] if rec["electric"] is not None else "无来源",
+        round(rec["hydrogen"] + rec["electric"], 3) if rec["hydrogen"] is not None else "无来源",
+        rec["tbox"] if rec["tbox"] is not None else "无来源",
+        rec["gps"] if rec["gps"] is not None else "无来源",
+    ])
+
+ws.insert_rows(1)
+ws.merge_cells("A1:L1")
+ws["A1"] = (
+    "1. 纯氢里程(km)、纯电里程(km)来源于广州平台统计，TBOX里程来源于车机上报的仪表盘里程统计，"
+    "GPS里程优先按两份7月GPS日明细汇总；测试车辆未命中时，回填7月最终考核源表中已确认的GPS实际里程\n"
+    "2. 广州平台总里程(km)=纯氢里程(km)+纯电里程(km)\n"
+    "3. 已确认广州平台暂未接入86辆普货车数据；无对应记录统一标记为“无来源”"
+)
+
+header_fill = PatternFill("solid", fgColor="1F4E78")
+platform_fill = PatternFill("solid", fgColor="DDEBF7")
+tbox_fill = PatternFill("solid", fgColor="E2F0D9")
+gps_fill = PatternFill("solid", fgColor="E4DFEC")
+test_fill = PatternFill("solid", fgColor="FFF2CC")
+missing_fill = PatternFill("solid", fgColor="F2F2F2")
+thin = Side(style="thin", color="B7B7B7")
+
+for cell in ws[2]:
+    cell.fill = header_fill
+    cell.font = Font(color="FFFFFF", bold=True)
+    cell.alignment = Alignment(horizontal="center", vertical="center", wrap_text=True)
+    cell.border = Border(bottom=Side(style="medium", color="FFFFFF"))
+
+for row in ws.iter_rows(min_row=3, max_row=ws.max_row):
+    is_test = row[6].value == "是"
+    for cell in row:
+        cell.border = Border(left=thin, right=thin, top=thin, bottom=thin)
+        cell.alignment = Alignment(horizontal="center", vertical="center")
+    row[2].alignment = Alignment(horizontal="left", vertical="center")
+    if is_test:
+        for cell in row:
+            cell.fill = test_fill
+    else:
+        for cell in row[7:10]:
+            cell.fill = platform_fill
+        row[10].fill = tbox_fill
+        row[11].fill = gps_fill
+        for cell in row[7:12]:
+            if cell.value == "无来源":
+                cell.fill = missing_fill
+                cell.font = Font(color="7F7F7F", italic=True)
+
+for row in range(3, ws.max_row + 1):
+    ws.cell(row, 4).number_format = "yyyy-mm-dd"
+    ws.cell(row, 5).number_format = "yyyy-mm-dd"
+    for col in range(8, 13):
+        ws.cell(row, col).number_format = "#,##0.0"
+
+widths = [8, 14, 34, 15, 15, 11, 14, 17, 17, 23, 17, 17]
+for col, width in enumerate(widths, 1):
+    ws.column_dimensions[get_column_letter(col)].width = width
+ws["A1"].fill = PatternFill("solid", fgColor="FFF2CC")
+ws["A1"].font = Font(color="7F6000", italic=True)
+ws["A1"].alignment = Alignment(horizontal="left", vertical="center", wrap_text=True)
+ws.row_dimensions[1].height = 58
+ws.row_dimensions[2].height = 36
+ws.freeze_panes = "A3"
+ws.auto_filter.ref = f"A2:L{ws.max_row}"
+ws.sheet_view.showGridLines = False
+ws.sheet_properties.pageSetUpPr.fitToPage = True
+ws.page_setup.orientation = "landscape"
+ws.page_setup.fitToWidth = 1
+ws.page_setup.fitToHeight = 0
+ws.print_title_rows = "1:2"
+
+OUTPUT.parent.mkdir(parents=True, exist_ok=True)
+wb.save(OUTPUT)
+
+print({
+    "output": str(OUTPUT),
+    "records": len(records),
+    "test_records": sum(r["test"] == "是" for r in records),
+    "platform_missing": sum(r["hydrogen"] is None for r in records),
+    "tbox_missing": sum(r["tbox"] is None for r in records),
+    "gps_missing": sum(r["gps"] is None for r in records),
+})
diff --git a/calc_engine.py b/calc_engine.py
index c4c9b2d..1dfcab7 100644
--- a/calc_engine.py
+++ b/calc_engine.py
@@ -2,15 +2,22 @@
 import openpyxl
 from collections import defaultdict
 
+# RULES: 18T用(1000/0/1000/1000), 非18T保持原逻辑(未达标不发, 结转补发=原奖励额)
 RULES = {
-    '交投40辆4.5T普货':    {'km': 3000, '奖': 150},
-    '交投190辆4.5T冷链车': {'km': 3000, '奖': 150},
-    '羚牛136辆4.5T冷链车': {'km': 5000, '奖': 260},
-    '恒运50辆4.5T普货':    {'km': 5000, '奖': 260},
-    '恒运50辆4.5T 普货':   {'km': 5000, '奖': 260},
-    '羚牛100辆18T':        {'km': 6000, '奖': 1000},
+    '交投40辆4.5T普货':    {'km': 3000, 'good': 150, 'bad': 0,   'carry': 150, 'other': 150},
+    '交投190辆4.5T冷链车': {'km': 3000, 'good': 150, 'bad': 0,   'carry': 150, 'other': 150},
+    '羚牛136辆4.5T冷链车': {'km': 5000, 'good': 260, 'bad': 0,   'carry': 260, 'other': 260},
+    '恒运50辆4.5T普货':    {'km': 5000, 'good': 260, 'bad': 0,   'carry': 260, 'other': 260},
+    '恒运50辆4.5T 普货':   {'km': 5000, 'good': 260, 'bad': 0,   'carry': 260, 'other': 260},
+    '羚牛100辆18T':        {'km': 6000, 'good': 1000, 'bad': 0, 'carry': 1000, 'other': 1000},
 }
-DAYS = {1: 31, 2: 28, 3: 31}
+
+# 2026年7月经业务确认的4辆18T执行下半年新政策：
+# 月目标5000公里，达标150元，未达标不发，结转/补发按150元基准。
+# 其余18T继续使用上面的原口径。
+JULY_SCHEME3_PLATES = {'粤AGR0772', '粤AGQ0178', '粤AGP6675', '粤AGP9757'}
+JULY_SCHEME3_RULE = {'km': 5000, 'good': 150, 'bad': 0, 'carry': 150, 'other': 150}
+DAYS = {1: 31, 2: 28, 3: 31, 4: 30, 5: 31, 6: 30, 7: 31, 8: 31, 9: 30, 10: 31, 11: 30, 12: 31}
 
 def read_file(fp, month):
     wb = openpyxl.load_workbook(fp, data_only=True)
@@ -23,13 +30,22 @@ def read_file(fp, month):
         r['应考核里程(km)'] = float(r.get('应考核里程(km)') or 0)
         r['实际行驶里程(km)'] = float(r.get('实际行驶里程(km)') or 0)
         r['月份'] = month
-        rule = RULES.get(r.get('考核目标', ''), {})
+        if month == 7 and str(r.get('车牌号') or '').strip().upper() in JULY_SCHEME3_PLATES:
+            rule = JULY_SCHEME3_RULE
+        else:
+            rule = RULES.get(r.get('考核目标', ''), {'km': 0, 'good': 0, 'bad': 0, 'carry': 0, 'other': 0})
         r['月度目标里程'] = rule.get('km', 0)
-        r['对应月奖励金额'] = rule.get('奖', 0)
+        r['奖励基准_达标'] = rule.get('good', 0)
+        r['奖励基准_未达标'] = rule.get('bad', 0)
+        r['奖励基准_结转'] = rule.get('carry', 0)
+        r['奖励基准_补发'] = rule.get('other', 0)
         t, a = r['应考核里程(km)'], r['实际行驶里程(km)']
         q = a >= t and t > 0
         r['是否达标'] = '达标' if q else '未达标'
-        r['奖金'] = r['对应月奖励金额'] * (r['考核天数'] / DAYS[month]) if q else 0
+        if q:
+            r['奖金'] = r['奖励基准_达标'] * (r['考核天数'] / DAYS[month])
+        else:
+            r['奖金'] = r['奖励基准_未达标'] * (r['考核天数'] / DAYS[month])
         r['多跑'] = max(0, a - t) if q else 0
         r['可结转'] = int(r['多跑'] // r['月度目标里程']) if r['月度目标里程'] > 0 and q else 0
         recs.append(r)
@@ -38,7 +54,8 @@ def read_file(fp, month):
 
 def grp(recs, month):
     gs = defaultdict(lambda: {'recs':[], '应考核':0, '实际':0, '奖金':0, '天数':0,
-                               '有达标':False, '目标km':0, '奖励额':0})
+                               '有达标':False, '目标km':0, '奖励达标':0, '奖励未达标':0,
+                               '奖励结转':0, '奖励补发':0})
     for r in recs:
         k = (r['车牌号'], r['销售经理'])
         g = gs[k]
@@ -49,7 +66,10 @@ def grp(recs, month):
         g['天数'] += r['考核天数']
         if r['是否达标'] == '达标': g['有达标'] = True
         g['目标km'] = max(g['目标km'], r['月度目标里程'])
-        g['奖励额'] = max(g['奖励额'], r['对应月奖励金额'])
+        g['奖励达标'] = max(g['奖励达标'], r['奖励基准_达标'])
+        g['奖励未达标'] = max(g['奖励未达标'], r['奖励基准_未达标'])
+        g['奖励结转'] = max(g['奖励结转'], r['奖励基准_结转'])
+        g['奖励补发'] = max(g['奖励补发'], r['奖励基准_补发'])
         g['部门'] = r['部门名称']
         g['销售'] = r['销售经理']
         g['车牌'] = r['车牌号']
@@ -61,153 +81,161 @@ def calc_jan_carryover(G1):
         for r in g['recs']:
             if r['是否达标'] == '达标': tc += r['可结转']
         g['可结转'] = tc
+        g['累计已发结转'] = 0
+        g['1月已发'] = g['有达标']  # 当月有达标 → 已发
+        g['cum_t'] = g['应考核']
+        g['cum_a'] = g['实际']
+        g['cum_q'] = g['实际'] >= g['应考核'] and g['应考核'] > 0
 
-def calc_feb(G1, G2):
-    feb_data = {'结转':[], '补发1月':[], '当月':[], '累计补发2月':[]}
-    for k, g2 in G2.items():
-        g1 = G1.get(k)
-        bf = g2['奖励额']
-        j_t = g1['应考核'] if g1 else 0; j_a = g1['实际'] if g1 else 0
-        j_q = g1['有达标'] if g1 else False; j_c = g1['可结转'] if g1 else 0
-        cum_t = j_t + g2['应考核']; cum_a = j_a + g2['实际']
+def calc_month(month, G_all, all_data, G_prev, G_cur):
+    """通用月计算: 结转/补发/累计补发/当月"""
+    prev_m = month - 1
+    days_m = DAYS[month]
+    data_m = {'结转': [], '当月': []}
+    # 向后兼容的key: 当月达标/当月未达标 用于18T区分
+    data_18t = {'当月达标': [], '当月未达标': []}
+    use_18t = True  # 统一使用拆分标签
+
+    for k, gm in G_cur.items():
+        gp = G_prev.get(k)
+
+        # 累计
+        cum_t = (gp['cum_t'] if gp else 0) + gm['应考核']
+        cum_a = (gp['cum_a'] if gp else 0) + gm['实际']
         cum_q = cum_a >= cum_t and cum_t > 0
 
+        # 结转: carry × 1 (整月, 不折算)
         carry = 0
-        if g1 and j_q and j_c >= 1:
-            carry = bf
-            feb_data['结转'].append({'车牌':k[0],'销售':g1['销售'],'部门':g2['部门'],'额':carry})
-        bp1 = 0
-        if g1 and not j_q and cum_q:
-            bp1 = g1['奖励额'] * (g1['天数'] / 31)
-            feb_data['补发1月'].append({'车牌':k[0],'销售':g1['销售'],'部门':g2['部门'],'额':bp1})
-        bonus2 = 0
-        if g2['有达标'] and carry == 0:
-            bonus2 = g2['奖金']
-            feb_data['当月'].append({'车牌':k[0],'销售':g2['销售'],'部门':g2['部门'],'额':bonus2})
-        cbp2 = 0
-        if not g2['有达标'] and carry == 0 and cum_q:
-            cbp2 = g2['奖励额'] * (g2['天数'] / 28)
-            feb_data['累计补发2月'].append({'车牌':k[0],'销售':g2['销售'],'部门':g2['部门'],'额':cbp2})
+        p_carry = gp['可结转'] if gp else 0
+        if gp and p_carry >= 1:
+            carry = gm['奖励结转']
+            data_m['结转'].append({'车牌': k[0], '销售': gp['销售'], '部门': gm['部门'], '额': carry})
 
-        g2['cum_t']=cum_t; g2['cum_a']=cum_a; g2['cum_q']=cum_q
-        g2['结转']=carry; g2['补发1月']=bp1
-        g2['补发1月对应']=g1['销售'] if g1 and bp1>0 else ''
-        g2['当月奖金']=bonus2; g2['累计补发2月']=cbp2; g2['结转占位']=carry>0
-        fc = sum(r['可结转'] for r in g2['recs'] if r['是否达标']=='达标')
-        jr = max(0, j_c - (1 if carry>0 else 0))
-        g2['可结转'] = fc + jr
-        g2['2月已发'] = carry>0 or bonus2>0 or cbp2>0
-        g2['1月已补发'] = bp1>0
+        # 当月奖金
+        is_18t = gm['奖励达标'] == 1000  # 18T特征
+        bonus = 0
+        if carry == 0 and gm['奖金'] > 0:
+            bonus = gm['奖金']
+            data_m['当月'].append({'车牌': k[0], '销售': gm['销售'], '部门': gm['部门'], '额': bonus})
 
-    # 补充：1月有结转但2月无考核记录的车 → 创建虚拟2月group并发放结转
-    for k, g1 in G1.items():
-        if k not in G2 and g1['可结转'] >= 1:
-            bf = g1['奖励额']
-            carry = bf
-            feb_data['结转'].append({'车牌':k[0],'销售':g1['销售'],'部门':g1['部门'],'额':carry})
-            # 创建虚拟2月group（考核里程=满月目标，实际=0）
-            G2[k] = {
-                'recs':[], '应考核':g1['目标km'], '实际':0, '奖金':0, '天数':DAYS[2],
-                '有达标':False, '目标km':g1['目标km'], '奖励额':g1['奖励额'],
-                '部门':g1['部门'], '销售':g1['销售'], '车牌':k[0],
-                'cum_t':g1['应考核']+g1['目标km'], 'cum_a':g1['实际'],
-                'cum_q': g1['实际'] >= g1['应考核']+g1['目标km'],
-                '结转':carry, '补发1月':0, '补发1月对应':'',
-                '当月奖金':0, '累计补发2月':0, '结转占位':True,
-                '可结转': max(0, g1['可结转'] - 1),
-                '2月已发':True, '1月已补发':False,
-                '虚拟':True,  # 标记为无考核记录
+        # 补发历史月: other × 折算
+        for back_m in range(1, month):
+            label = f'补发{back_m}月'
+            if label not in data_m:
+                data_m[label] = []
+            g_bm = G_all[back_m].get(k)
+            if not g_bm:
+                continue
+            bm_paid = g_bm.get(f'{back_m}月已发', g_bm['有达标'])
+            if bm_paid:
+                continue
+            bm_paid_later = False
+            for later_m in range(back_m + 1, month):
+                gl = G_all[later_m].get(k)
+                if gl and gl.get(f'补发{back_m}月', 0) > 0:
+                    bm_paid_later = True
+                    break
+            if bm_paid_later:
+                continue
+            if cum_q:
+                bj = g_bm['奖励补发'] * (g_bm['天数'] / DAYS[back_m])
+                data_m[label].append({'车牌': k[0], '销售': g_bm['销售'], '部门': gm['部门'], '额': bj})
+                gm[f'补发{back_m}月'] = bj
+
+        # 累计补发当月: other × 折算
+        label_cbp = f'累计补发{month}月'
+        if label_cbp not in data_m:
+            data_m[label_cbp] = []
+        cbp = 0
+        if not gm['有达标'] and carry == 0 and cum_q:
+            cbp = gm['奖励补发'] * (gm['天数'] / days_m)
+            data_m[label_cbp].append({'车牌': k[0], '销售': gm['销售'], '部门': gm['部门'], '额': cbp})
+
+        gm['cum_t'] = cum_t
+        gm['cum_a'] = cum_a
+        gm['cum_q'] = cum_q
+        gm['结转'] = carry
+        gm['当月奖金'] = bonus
+        gm['累计补发当月'] = cbp
+        gm[f'{month}月已发'] = carry > 0 or gm['奖金'] > 0 or cbp > 0
+
+        # 更新可结转
+        cum_over = max(0, cum_a - cum_t) if cum_q else 0
+        earned = int(cum_over // gm['目标km']) if gm['目标km'] > 0 else 0
+        prev_used = gp['累计已发结转'] if gp else 0
+        cur_used = prev_used + (1 if carry > 0 else 0)
+        gm['累计已发结转'] = cur_used
+        gm['可结转'] = max(0, earned - cur_used)
+
+    # 补充: 上月有结转但本月无考核记录的车
+    for k, gp in G_prev.items():
+        if k not in G_cur and gp.get('可结转', 0) >= 1:
+            carry = gp['奖励结转']
+            data_m['结转'].append({'车牌': k[0], '销售': gp['销售'], '部门': gp['部门'], '额': carry})
+            G_cur[k] = {
+                'recs': [], '应考核': gp['目标km'], '实际': 0, '奖金': 0, '天数': days_m,
+                '有达标': False, '目标km': gp['目标km'],
+                '奖励达标': gp['奖励达标'], '奖励未达标': gp['奖励未达标'],
+                '奖励结转': gp['奖励结转'], '奖励补发': gp['奖励补发'],
+                '部门': gp['部门'], '销售': gp['销售'], '车牌': k[0],
+                'cum_t': gp.get('cum_t', 0) + gp['目标km'], 'cum_a': gp.get('cum_a', 0),
+                'cum_q': False,
+                '结转': carry, '当月奖金': 0,
+                f'{month}月已发': True,
+                '可结转': max(0, gp['可结转'] - 1),
+                '累计已发结转': gp.get('累计已发结转', 0) + 1,
+                '虚拟': True,
             }
-    return feb_data
 
-def calc_mar(G1, G2, G3, feb_data):
-    mar_data = {'结转':[], '补发1月':[], '补发2月':[], '当月':[], '累计补发3月':[]}
-    for k, g3 in G3.items():
-        g1 = G1.get(k); g2 = G2.get(k)
-        bf = g3['奖励额']
-        j_t=g1['应考核'] if g1 else 0; j_a=g1['实际'] if g1 else 0
-        j_q=g1['有达标'] if g1 else False; j_paid=(g1['奖金']>0) if g1 else False
-        f_t=g2['应考核'] if g2 else 0; f_a=g2['实际'] if g2 else 0
-        f_paid=g2['2月已发'] if g2 else False; f_carry=g2['可结转'] if g2 else 0
-        if g1 and j_q: j_paid=True
-        if g2 and g2.get('1月已补发',False): j_paid=True
-        cum_t=j_t+f_t+g3['应考核']; cum_a=j_a+f_a+g3['实际']
-        cum_q=cum_a>=cum_t and cum_t>0
+    all_data[month] = data_m
+    return data_m, G_cur
 
-        carry=0
-        if g2 and f_carry>=1:
-            carry=bf; mar_data['结转'].append({'车牌':k[0],'销售':g2['销售'],'部门':g3['部门'],'额':carry})
-        bj=0
-        if g1 and not j_paid and cum_q:
-            bj=g1['奖励额']*(g1['天数']/31); mar_data['补发1月'].append({'车牌':k[0],'销售':g1['销售'],'部门':g3['部门'],'额':bj})
-        bf2=0
-        if g2 and not f_paid and cum_q:
-            bf2=g2['奖励额']*(g2['天数']/28); mar_data['补发2月'].append({'车牌':k[0],'销售':g2['销售'],'部门':g3['部门'],'额':bf2})
-        bonus3=0
-        if g3['有达标'] and carry==0:
-            bonus3=g3['奖金']; mar_data['当月'].append({'车牌':k[0],'销售':g3['销售'],'部门':g3['部门'],'额':bonus3})
-        cbp3=0
-        if not g3['有达标'] and carry==0 and cum_q:
-            cbp3=g3['奖励额']*(g3['天数']/31); mar_data['累计补发3月'].append({'车牌':k[0],'销售':g3['销售'],'部门':g3['部门'],'额':cbp3})
 
-        g3['cum_t']=cum_t; g3['cum_a']=cum_a; g3['cum_q']=cum_q
-        g3['结转']=carry; g3['补发1月']=bj; g3['补发1月对应']=g1['销售'] if g1 and bj>0 else ''
-        g3['补发2月']=bf2; g3['补发2月对应']=g2['销售'] if g2 and bf2>0 else ''
-        g3['当月奖金']=bonus3; g3['累计补发3月']=cbp3; g3['结转占位']=carry>0
-
-    # 补充：2月有结转但3月无考核记录的车
-    for k, g2 in G2.items():
-        if k not in G3 and g2.get('可结转', 0) >= 1:
-            bf = g2['奖励额']
-            carry = bf
-            mar_data['结转'].append({'车牌':k[0],'销售':g2['销售'],'部门':g2['部门'],'额':carry})
-            G3[k] = {
-                'recs':[], '应考核':g2['目标km'], '实际':0, '奖金':0, '天数':DAYS[3],
-                '有达标':False, '目标km':g2['目标km'], '奖励额':g2['奖励额'],
-                '部门':g2['部门'], '销售':g2['销售'], '车牌':k[0],
-                'cum_t':(G1.get(k,{}).get('应考核',0))+g2['应考核']+g2['目标km'],
-                'cum_a':(G1.get(k,{}).get('实际',0))+g2['实际'],
-                'cum_q':False,
-                '结转':carry, '补发1月':0, '补发1月对应':'',
-                '补发2月':0, '补发2月对应':'',
-                '当月奖金':0, '累计补发3月':0, '结转占位':True,
-                '虚拟':True,
-            }
-    return mar_data
-
-def collect_vehicle_payments(G, feb_data, mar_data):
+def collect_vehicle_payments(G, all_data):
+    """收集所有月的付款记录"""
     payments = defaultdict(list)
     info = {}
+
+    # 1月
     for k, g in G[1].items():
         plate = k[0]
         if not info.get(plate):
             first = g['recs'][0]
-            info[plate] = {'考核目标':first.get('考核目标',''),'月度奖励':g['奖励额'],'目标km':g['目标km']}
+            info[plate] = {'考核目标': first.get('考核目标', ''), '月度奖励': g['奖励达标'], '目标km': g['目标km']}
         if g['奖金'] > 0:
-            payments[plate].append({'结算月':1,'对应考核月':1,'业务员':g['销售'],'金额':g['奖金'],'类型':'当月达标','部门':g['部门']})
+            payments[plate].append({'结算月': 1, '对应考核月': 1, '业务员': g['销售'],
+                                     '金额': g['奖金'], '类型': '当月达标', '部门': g['部门']})
 
-    cat_map = {
-        ('结转',2):2, ('补发1月',2):1, ('当月',2):2, ('累计补发2月',2):2,
-        ('结转',3):3, ('补发1月',3):1, ('补发2月',3):2, ('当月',3):3, ('累计补发3月',3):3,
-    }
-    for month, mdata in [(2, feb_data), (3, mar_data)]:
+    # 2月-7月
+    cat_map = {}
+    for month in range(2, 8):
+        cat_map[('结转', month)] = month
+        for bm in range(1, month):
+            cat_map[(f'补发{bm}月', month)] = bm
+        cat_map[('当月', month)] = month
+        cat_map[(f'累计补发{month}月', month)] = month
+
+    for month in range(2, 8):
+        mdata = all_data.get(month, {})
         for cat, dl in mdata.items():
             assess_month = cat_map.get((cat, month), month)
             for d in dl:
-                payments[d['车牌']].append({'结算月':month,'对应考核月':assess_month,
-                    '业务员':d['销售'],'金额':d['额'],'类型':cat,'部门':d['部门']})
+                payments[d['车牌']].append({'结算月': month, '对应考核月': assess_month,
+                    '业务员': d['销售'], '金额': d['额'], '类型': cat, '部门': d['部门']})
 
     for m in G:
         for k, g in G[m].items():
             if not info.get(k[0]):
-                first = g['recs'][0]
-                info[k[0]] = {'考核目标':first.get('考核目标',''),'月度奖励':g['奖励额'],'目标km':g['目标km']}
+                first = g['recs'][0] if g['recs'] else None
+                if first:
+                    info[k[0]] = {'考核目标': first.get('考核目标', ''), '月度奖励': g['奖励达标'], '目标km': g['目标km']}
 
     for plate in payments:
         records = sorted(payments[plate], key=lambda x: (x['对应考核月'], x['结算月']))
         for i, r in enumerate(records): r['期数'] = i + 1
     return payments, info
 
+
 # (归属公司+车型) → 考核目标 映射，用于补全无考核记录的车辆
 VEHICLE_TARGET_MAP = {
     ('广州开发区交投氢能运营管理有限公司', '4.5吨冷链车'): ('交投190辆4.5T冷链车', 3000, 150),
@@ -218,6 +246,7 @@ VEHICLE_TARGET_MAP = {
     ('现代氢能科技（广州）有限公司', '4.5吨货车'): ('恒运50辆4.5T普货', 5000, 260),
 }
 
+
 def read_loss_data(month):
     """读取亏损表，返回 {客户名称: '是'/'否'} 字典。无亏损表返回None"""
     import os
@@ -229,8 +258,6 @@ def read_loss_data(month):
     ws = wb[wb.sheetnames[0]]
     h = [c.value for c in next(ws.iter_rows(min_row=1, max_row=1))]
 
-    # 1月: 列名"项目","1月是否亏损"
-    # 2月: 列名"客户名称","是否亏损"
     client_col = None; loss_col = None
     for i, col_name in enumerate(h):
         if col_name and ('项目' in str(col_name) or '客户' in str(col_name)):
@@ -251,6 +278,7 @@ def read_loss_data(month):
     wb.close()
     return result
 
+
 def get_vehicle_client_map(D):
     """从考核源数据构建 {车牌号: 客户名称} 映射（取最新月的客户名）"""
     plate_client = {}
@@ -259,6 +287,7 @@ def get_vehicle_client_map(D):
             plate_client[r['车牌号']] = r.get('客户名称', '')
     return plate_client
 
+
 def read_master_vehicles(fp='里程任务考核_Q1汇总.xlsx'):
     """从现有Q1汇总文件读取全量车辆台账"""
     wb = openpyxl.load_workbook(fp, data_only=True)
@@ -268,7 +297,7 @@ def read_master_vehicles(fp='里程任务考核_Q1汇总.xlsx'):
     for row in ws.iter_rows(min_row=2, values_only=True):
         r = dict(zip(h, row))
         if r.get('车牌号'):
-            vehicles.append({'车牌号': r['车牌号'], '车架号': r.get('车架号',''),
-                             '归属公司': r.get('归属公司',''), '车型确定': r.get('车型','')})
+            vehicles.append({'车牌号': r['车牌号'], '车架号': r.get('车架号', ''),
+                             '归属公司': r.get('归属公司', ''), '车型确定': r.get('车型', '')})
     wb.close()
     return vehicles
diff --git a/carbon_platform_peer_discussion_questions.md b/carbon_platform_peer_discussion_questions.md
new file mode 100644
index 0000000..59dd533
--- /dev/null
+++ b/carbon_platform_peer_discussion_questions.md
@@ -0,0 +1,549 @@
+# 初次建设碳平台：同行交流问题清单
+
+> 适用场景：企业第一次建设碳管理 / 碳资产 / 碳核算平台时，用于和同行、供应商、咨询机构、碳管理团队交流。
+
+---
+
+## 一、先明确：碳平台到底服务谁？
+
+1. 你们做碳平台的第一目标是什么？
+   - 政府监管？
+   - 集团管理？
+   - 客户披露？
+   - 碳交易履约？
+   - ESG 报告？
+
+2. 平台一期最先做的是哪块？
+   - 碳盘查
+   - 碳核算
+   - 碳资产
+   - 碳交易
+   - 产品碳足迹
+   - 能源管理
+
+3. 你们一开始有没有做“大而全”？后来哪些模块发现其实用不上？
+
+4. 平台最终是给谁每天使用？
+   - 安环部
+   - 能源管理部
+   - 财务部
+   - 生产部门
+   - 集团总部
+   - 第三方核查机构
+
+5. 你们怎么定义平台成功？
+   - 减少人工填报？
+   - 提升核算准确性？
+   - 支撑履约？
+   - 形成减排决策？
+
+---
+
+## 二、碳核算边界怎么定？
+
+这是最容易前期没想清、后期返工的地方。
+
+1. 组织边界怎么划？
+   - 按法人？
+   - 按厂区？
+   - 按项目？
+   - 按车队？
+   - 按站点？
+   - 按集团层级？
+
+2. 运营边界怎么划？
+   - 范围一做到什么程度？
+   - 范围二做到什么程度？
+   - 范围三是否纳入？
+
+3. 哪些排放源必须纳入？
+   - 电
+   - 气
+   - 煤
+   - 油
+   - 蒸汽
+   - 制冷剂
+   - 运输
+   - 外购服务
+   - 员工通勤
+
+4. 对于多业务混合场景，怎么拆分碳排放？
+   - 一个站点服务多个客户，怎么分摊？
+   - 一个车辆服务多个项目，怎么分摊？
+   - 一个能源账单对应多个业务单元，怎么分摊？
+
+5. 历史数据从哪一年开始补？
+   - 只做当前年度？
+   - 补 3 年？
+   - 补 5 年？
+   - 是否需要和历史报告口径一致？
+
+---
+
+## 三、数据来源和数据质量问题
+
+碳平台本质上很大程度是一个数据平台，而不是单纯报表系统。
+
+1. 你们碳数据主要来自哪里？
+   - ERP
+   - MES
+   - SCADA
+   - 电表系统
+   - 能源系统
+   - 财务系统
+   - 人工 Excel
+   - 第三方平台
+
+2. 哪些数据最难自动获取？
+   - 电费
+   - 燃气
+   - 柴油
+   - 蒸汽
+   - 原材料
+   - 运输里程
+   - 供应商数据
+
+3. 你们有没有做原始凭证管理？
+   - 发票
+   - 账单
+   - 抄表照片
+   - 采购单
+   - 加油记录
+   - 加氢记录
+
+4. 数据质量怎么校验？
+   - 缺失值
+   - 异常值
+   - 重复数据
+   - 跨月修正
+   - 人工调整留痕
+
+5. 人工填报和系统采集冲突时，以谁为准？
+
+6. 有没有“数据质量评分”？
+   - 自动采集数据是否优于人工填报？
+   - 带凭证数据是否优于无凭证数据？
+
+7. 月度数据锁定机制怎么做？
+   - 锁定后还能不能修改？
+   - 修改是否需要审批？
+   - 是否保留修改前后的版本？
+
+---
+
+## 四、核算模型和排放因子怎么维护？
+
+1. 你们核算方法是内置固定模板，还是可配置公式？
+
+2. 排放因子怎么管理？
+   - 国家因子
+   - 省级因子
+   - 行业因子
+   - 企业自定义因子
+
+3. 排放因子优先级怎么定？
+
+4. 排放因子版本如何追溯？
+   - 2024 年用了 A 因子，2025 年用了 B 因子，历史结果是否重算？
+
+5. 活动数据和排放因子是否分开存？
+   - 只存最终碳排放结果，还是保留计算过程？
+
+6. 核算公式有没有审批发布流程？
+
+7. 不同行业是否做了不同模板？
+   - 电力
+   - 钢铁
+   - 水泥
+   - 铝
+   - 化工
+   - 交通运输
+   - 园区
+
+8. 平台是否支持核算结果解释？
+   - 能不能点开一条排放数据，看到：来源数据 × 排放因子 × 公式 = 结果？
+
+---
+
+## 五、MRV：监测、报告、核查怎么做？
+
+1. 平台是否覆盖完整 MRV 流程？
+   - Monitoring：监测
+   - Reporting：报告
+   - Verification：核查
+
+2. 第三方核查机构怎么使用平台？
+   - 给账号直接进系统看？
+   - 导出材料给核查机构？
+   - 是否有专门的核查视图？
+
+3. 核查问题如何闭环？
+   - 核查意见
+   - 整改记录
+   - 补充材料
+   - 版本留痕
+
+4. 是否支持自动生成排放报告？
+
+5. 报告模板是否按行业、地区、监管要求区分？
+
+6. 是否有审计轨迹？
+   - 谁填的？
+   - 谁改的？
+   - 谁审的？
+   - 什么时候提交的？
+
+7. 是否做了填报任务管理？
+   - 每月提醒
+   - 逾期提醒
+   - 部门催办
+   - 审批流
+
+---
+
+## 六、碳资产和碳交易怎么设计？
+
+如果涉及全国碳市场或地方碳市场，这块要提前问清楚。
+
+1. 平台是否管理配额？
+   - 初始配额
+   - 调整配额
+   - 已用配额
+   - 剩余配额
+   - 履约缺口
+
+2. 是否支持 CCER 或其他减排量资产管理？
+
+3. 是否和交易系统打通？
+   - 直接交易？
+   - 只做台账？
+   - 只做履约测算？
+
+4. 碳价怎么维护？
+   - 手工录入
+   - 接口获取
+   - 财务确认
+
+5. 是否做履约预测？
+   - 当前排放趋势下，年底预计缺多少配额？
+   - 是否需要提前采购？
+
+6. 是否支持碳资产成本测算？
+   - 买配额成本
+   - 减排改造成本
+   - 不同策略对比
+
+7. 财务上怎么入账？
+   - 碳配额
+   - 碳资产
+   - 履约成本
+   - 减排收益
+
+---
+
+## 七、减排分析和业务决策怎么落地？
+
+很多碳平台最后容易变成“填报系统”，所以要重点问同行怎么避免。
+
+1. 平台有没有做减排潜力分析？
+
+2. 是否能按不同维度拆分碳排放？
+   - 组织
+   - 站点
+   - 车辆
+   - 设备
+   - 产品
+   - 客户
+   - 线路
+
+3. 是否能看单位指标？
+   - 吨产品碳排
+   - 百公里碳排
+   - 单车碳排
+   - 单位营收碳排
+   - 单位能耗碳排
+
+4. 是否能识别异常排放？
+   - 某站点本月电耗异常
+   - 某车辆能耗异常
+   - 某条线路碳排异常
+
+5. 是否能支撑减排项目管理？
+   - 项目立项
+   - 预计减排量
+   - 实际减排量
+   - 投资回收期
+
+6. 有没有做情景预测？
+   - 电价变化
+   - 绿电比例提升
+   - 车辆替换
+   - 工艺改造
+   - 能源结构调整
+
+7. 平台输出的数据，业务部门真的会用吗？
+   - 用于经营分析？
+   - 用于客户报价？
+   - 用于 ESG 披露？
+   - 还是只给领导看大屏？
+
+---
+
+## 八、产品碳足迹是否要做？
+
+产品碳足迹一期不一定要上，但要提前了解是否存在外部压力。
+
+1. 你们有没有做产品碳足迹？
+
+2. 如果做，是按什么标准？
+   - ISO 14067
+   - PAS 2050
+   - 客户自定义规则
+   - 行业规则
+
+3. 生命周期边界做到哪里？
+   - cradle-to-gate
+   - cradle-to-grave
+   - 只做到企业内部生产环节
+
+4. 原材料供应商数据怎么获取？
+
+5. 供应商不给数据时怎么处理？
+   - 用行业平均值
+   - 用数据库因子
+   - 要求补充证明
+   - 先人工估算
+
+6. 产品碳足迹和企业碳核算是否共用一套数据底座？
+
+7. 有没有客户、出口、CBAM、ESG 披露等实际压力？
+
+---
+
+## 九、平台架构和系统集成问题
+
+可以结合现有业务系统、数据中台、车辆平台、能源账单系统来交流。
+
+1. 平台是独立建设，还是接入现有数据中台？
+
+2. 主数据怎么定？
+   - 组织
+   - 厂区
+   - 站点
+   - 车辆
+   - 设备
+   - 能源品类
+   - 供应商
+   - 客户
+
+3. 是否需要和这些系统打通？
+   - ERP
+   - 财务系统
+   - 电表系统
+   - 能源管理系统
+   - 车辆 TBox
+   - 加氢站系统
+   - 合同系统
+   - 采购系统
+
+4. 碳平台的数据粒度是什么？
+   - 实时
+   - 日级
+   - 月级
+   - 年度
+
+5. 对于碳核算来说，实时数据有没有必要？
+   - 哪些场景需要实时？
+   - 哪些场景只需要月度？
+
+6. 是否做数据血缘？
+   - 一个碳排结果能不能追溯到原始业务单据？
+
+7. 权限怎么设计？
+   - 集团
+   - 子公司
+   - 厂区
+   - 部门
+   - 第三方核查机构
+
+8. 是否支持多租户？
+   - 如果未来给多个客户、园区、站点使用，架构能不能撑住？
+
+---
+
+## 十、实施踩坑问题
+
+这些问题最容易问出真实经验。
+
+1. 第一期最不该做的功能是什么？
+
+2. 最大的坑是技术问题、数据问题，还是组织协同问题？
+
+3. 哪些数据一开始以为有，后来发现没有？
+
+4. 哪些数据口径最容易产生争议？
+   - 财务账单
+   - 抄表数
+   - 生产系统数
+   - 业务系统数
+
+5. 平台上线后，谁负责持续维护？
+   - 排放因子
+   - 核算公式
+   - 组织架构
+   - 数据质量
+   - 报告模板
+
+6. 有没有专门的碳管理岗位？
+   - 安环负责？
+   - 财务负责？
+   - 能源部门负责？
+   - 信息化负责？
+   - 多部门共同维护？
+
+7. 你们上线用了多久？
+   - 试点范围多大？
+   - 先选一个厂区？
+   - 先选一个园区？
+   - 还是集团全铺？
+
+8. 系统上线后，人工 Excel 有没有真的减少？
+
+9. 做平台前有没有先做碳盘查咨询？
+   - 咨询成果怎么转成系统需求？
+
+10. 平台建设是 IT 主导，还是碳管理部门主导？最终谁说了算？
+
+---
+
+## 十一、结合车辆 / 加氢 / 能源业务的重点追问方向
+
+如果企业业务涉及车辆运营、加氢站、能源账单、TBox、物流轨迹，可以重点问以下方向。
+
+### 1. 交通 / 车队碳核算
+
+1. 车辆碳排是按燃料消耗核算，还是按里程和车型因子估算？
+
+2. 新能源 / 氢能车辆怎么计算减排量？
+
+3. 车辆、司机、客户、线路之间的碳排怎么分摊？
+
+4. 空驶、异常行驶、运营行驶是否区分？
+
+5. 有没有把 TBox 里程、加氢订单、ETC、物流单结合起来做碳排核算？
+
+6. 车辆维度是否支持以下指标？
+   - 单车碳排
+   - 百公里碳排
+   - 单趟运输碳排
+   - 客户维度碳排
+   - 线路维度碳排
+
+### 2. 加氢站 / 能源侧碳数据
+
+1. 加氢站的电耗、氢源、压缩、储运环节碳排怎么核算？
+
+2. 不同氢源的排放因子怎么维护？
+   - 灰氢
+   - 蓝氢
+   - 绿氢
+   - 外购氢
+
+3. 站点级碳排是否能分摊到车辆或客户？
+
+4. 是否记录每笔加氢订单对应的碳因子？
+
+5. 氢气采购、运输、站内损耗是否纳入？
+
+6. 站点电耗与加氢量之间是否做能效分析？
+
+### 3. 碳平台与业务系统融合
+
+1. 碳平台是单独填报，还是从业务系统自动取数？
+
+2. 如果业务系统数据不完整，碳平台是反向推动业务系统补数据，还是允许人工补录？
+
+3. 碳数据能不能回流业务？
+   - 客户账单
+   - 绿色运力证明
+   - ESG 报告
+   - 项目报价
+   - 客户披露材料
+
+4. 有没有把碳排放作为业务 KPI？
+
+5. 碳平台是否能和车辆资产、合同、订单、账单系统联动？
+
+---
+
+## 十二、可直接带去交流的高质量问题清单
+
+下面这组问题适合在同行交流会上直接使用。
+
+1. 你们碳平台一期的最小可用范围是什么？
+
+2. 组织边界、运营边界、核算边界是谁定的？
+
+3. 数据主要来自系统接口还是人工填报？比例大概多少？
+
+4. 活动数据、排放因子、核算公式是否都能版本追溯？
+
+5. 核算结果能否追溯到原始凭证？
+
+6. 月度数据有没有锁定、审批、修改留痕机制？
+
+7. 第三方核查机构怎么使用平台？
+
+8. 平台是否支持全国碳市场履约和配额管理？
+
+9. 是否做减排预测，而不仅是排放统计？
+
+10. 上线后，哪些功能使用频率最高？哪些功能基本没人用？
+
+11. 最大的实施难点是数据、口径、组织协同，还是系统集成？
+
+12. 如果重做一次，你们第一期会砍掉哪些功能？
+
+13. 你们的排放因子、公式、报告模板由谁维护？
+
+14. 碳平台上线后，有没有真正减少人工 Excel？
+
+15. 平台输出的数据是否参与经营决策？还是主要用于报告和展示？
+
+---
+
+## 十三、交流时建议关注的判断标准
+
+和同行交流时，不建议只问“有什么功能”，而要重点判断这些问题：
+
+1. 数据是否可追溯
+   - 能否从碳排结果追到原始单据、原始读数、原始凭证。
+
+2. 口径是否可解释
+   - 每个排放结果是否能说明怎么算出来的。
+
+3. 过程是否可核查
+   - 是否有审批、留痕、版本、附件、核查意见闭环。
+
+4. 结果是否可使用
+   - 是否真正服务履约、减排、经营、客户披露，而不是只做大屏。
+
+5. 平台是否可持续维护
+   - 因子、公式、组织、模板、权限、数据源变化后，是否能低成本维护。
+
+---
+
+## 十四、总结建议
+
+第一次做碳平台，不要先问：
+
+> 系统有什么功能？
+
+而要先问：
+
+> 数据从哪里来？口径谁确认？结果给谁用？能不能被核查？上线后谁维护？
+
+碳平台的核心不是大屏，而是形成一套：
+
+> 可追溯、可核查、可解释、可持续维护的数据闭环。
+
diff --git a/excel_writer.py b/excel_writer.py
index a830949..12fa856 100644
--- a/excel_writer.py
+++ b/excel_writer.py
@@ -61,22 +61,32 @@ def write_total(ws, rn, month, all_data):
 
 def write_rules_sheet(wb):
     ws = wb.active; ws.title='考核奖励规则'
-    WH(ws,['考核目标','月度目标里程(km)','奖励金额(元)'])
-    for i,(n,km,b) in enumerate([('交投40辆4.5T普货',3000,150),('交投190辆4.5T冷链车',3000,150),
-        ('羚牛136辆4.5T冷链车',5000,260),('恒运50辆4.5T普货',5000,260),('羚牛100辆18T',6000,1000)],2):
-        WR(ws,i,[n,km,b])
+    WH(ws,['考核目标','月度目标里程(km)','当月达标','当月未达标','结转','补发/累计补发'])
+    for i,(n,km,g,b,c,o,is18t) in enumerate([
+        ('交投40辆4.5T普货',3000,150,0,150,150,False),
+        ('交投190辆4.5T冷链车',3000,150,0,150,150,False),
+        ('羚牛136辆4.5T冷链车',5000,260,0,260,260,False),
+        ('恒运50辆4.5T普货',5000,260,0,260,260,False),
+        ('羚牛100辆18T',6000,1000,0,1000,1000,True),
+        ('2026年7月指定4辆18T',5000,150,0,150,150,True),
+    ],2):
+        WR(ws,i,[n,km,
+            f'{g}元×天数折算',
+            f'{b}元×天数折算' if is18t else '不发',
+            f'{c}元整月',
+            f'{o}元×折算'])
     AW(ws)
 
 def write_detail_sheet(wb, records, month):
     ws = wb.create_sheet(f'里程明细{month}月')
-    WH(ws,['车牌号','部门名称','销售经理','客户名称','合同编号','考核目标','月度目标里程','对应月奖励金额',
+    WH(ws,['车牌号','部门名称','销售经理','客户名称','合同编号','考核目标','月度目标里程','月奖励金额',
             '考核天数',f'{month}月应考核里程',f'{month}月实际行驶里程','完成率',f'{month}月是否达标',
             f'{month}月奖金(天数折算)','多跑里程','可结转月数','考核状态'])
     rn=2
     for r in sorted(records, key=lambda x: (x['销售经理'],x['车牌号'])):
         t,a = r['应考核里程(km)'], r['实际行驶里程(km)']
         WR(ws,rn,[r['车牌号'],r['部门名称'],r['销售经理'],r.get('客户名称',''),r.get('合同编号',''),
-            r.get('考核目标',''),r['月度目标里程'],r['对应月奖励金额'],r['考核天数'],
+            r.get('考核目标',''),r['月度目标里程'],r['奖励基准_达标'],r['考核天数'],
             R(t),R(a),R(a/t,4) if t>0 else 0,r['是否达标'],R(r['奖金']),R(r['多跑']),r['可结转'],
             r.get('考核状态','')]); rn+=1
     AW(ws)
@@ -162,6 +172,41 @@ def write_calc_process_mar(wb, G1, G2, G3, feb_data):
             c5,R(cbp3) if cbp3>0 else 0,R(total) if total>0 else 0]); rn+=1
     AW(ws)
 
+def write_calc_process_generic(wb, month, G_all, all_data):
+    """通用计算过程sheet，支持1-6月"""
+    G_cur = G_all[month]
+    headers = ['车牌号','销售经理','部门',
+               f'{month}月记录数',f'{month}月总天数',f'{month}月应考核',f'{month}月实际',f'{month}月有达标',f'{month}月奖金',
+               '累计应考核','累计实际','累计达标','可结转','结转发放']
+    headers += [f'补发{m}月' for m in range(1, month)]
+    headers += ['当月奖金', f'累计补发{month}月', f'{month}月发放合计']
+
+    ws = wb.create_sheet(f'{month}月计算过程')
+    WH(ws, headers)
+    rn = 2
+    for k in sorted(G_cur.keys(), key=lambda x: (x[1], x[0])):
+        g = G_cur[k]
+        if g.get('虚拟'):
+            continue
+        cum_q_str = '达标' if g.get('cum_q') else '未达标'
+        dq_str = '是' if g['有达标'] else '否'
+        carry = g.get('结转', 0)
+        bonus = g.get('当月奖金', 0)
+        cbp = g.get('累计补发当月', 0)
+        back_pays = [g.get(f'补发{m}', 0) for m in range(1, month)]
+        total = carry + bonus + cbp + sum(back_pays)
+
+        row = [k[0], k[1], g['部门'],
+               len(g['recs']), g['天数'], R(g['应考核']), R(g['实际']), dq_str, R(g['奖金']),
+               R(g.get('cum_t', 0)), R(g.get('cum_a', 0)), cum_q_str,
+               g.get('可结转', 0), R(carry)]
+        row += [R(bp) for bp in back_pays]
+        row += [R(bonus), R(cbp), R(total)]
+        WR(ws, rn, row)
+        rn += 1
+    AW(ws)
+
+
 def write_summary_jan(wb, records, loss_data=None, plate_client=None):
     ws = wb.create_sheet('1月汇总')
     jan_dl=[{'车牌':r['车牌号'],'销售':r['销售经理'],'部门':r['部门名称'],'额':r['奖金']} for r in records if r['是否达标']=='达标']
@@ -452,7 +497,7 @@ def write_salesperson_sheet(wb, person, dept, settle_month, D, G, month_data, ve
         mkm = g_cur['目标km']
 
         # 第1行：车辆信息 + 各月里程/目标
-        row = [plate, first.get('考核目标',''), g_cur['奖励额']]
+        row = [plate, first.get('考核目标',''), g_cur['奖励达标']]
         cum_t = 0; cum_a = 0
         for m in range(1, settle_month+1):
             gm = G.get(m, {}).get((plate, person))
diff --git a/fix_july_intervention_style.py b/fix_july_intervention_style.py
new file mode 100644
index 0000000..bf59daf
--- /dev/null
+++ b/fix_july_intervention_style.py
@@ -0,0 +1,147 @@
+#!/usr/bin/env python3
+"""复刻6月干预说明配色，并将达标分类改为一行一人。"""
+from copy import copy
+from collections import defaultdict
+from pathlib import Path
+import openpyxl
+
+BASE = Path('/Users/kkfluous/Downloads')
+JUL = BASE / '7月核算结果/租赁任务考核_2026年7月_干预情况说明.xlsx'
+JUN = BASE / '6月核算结果/租赁任务考核_2026年6月_干预情况说明.xlsx'
+SRC = BASE / '7月核算结果/租赁任务考核_2026年7月.xlsx'
+BAK = BASE / '7月核算结果/租赁任务考核_2026年7月_原版备份.xlsx'
+TEST = Path('/Users/kkfluous/Library/Containers/com.tencent.xinWeChat/Data/Documents/xwechat_files/wxid_1704407055112_3af8/temp/RWTemp/2026-08/4340099467fbf08b4c6583c49adba4ee/7月测试车辆明细.xlsx')
+
+def recs(path):
+    wb = openpyxl.load_workbook(path, data_only=True)
+    ws = wb['业务考核视图']
+    h = [c.value for c in ws[1]]
+    out = [dict(zip(h, r)) for r in ws.iter_rows(min_row=2, values_only=True) if r[0]]
+    wb.close()
+    return out
+
+def clone_style(src, dst):
+    dst.font = copy(src.font)
+    dst.fill = copy(src.fill)
+    dst.border = copy(src.border)
+    dst.alignment = copy(src.alignment)
+    dst.protection = copy(src.protection)
+    dst.number_format = src.number_format
+
+def style_row(ws, row, cols, ref_ws, ref_row, ref_cols=None):
+    ref_cols = ref_cols or cols
+    for c, rc in zip(cols, ref_cols):
+        clone_style(ref_ws.cell(ref_row, rc), ws.cell(row, c))
+
+new = recs(SRC)
+old = recs(BAK)
+old_map = {}
+for r in old:
+    old_map[r['车牌号']] = r
+
+twb = openpyxl.load_workbook(TEST, data_only=True)
+test_plates = {str(r[0]).strip() for r in twb.active.iter_rows(min_row=2, values_only=True) if r[0]}
+test_plates.discard('粤AG37316')
+twb.close()
+intervention = [r for r in new if r['车牌号'] in test_plates]
+flip = []
+for r in intervention:
+    old_r = old_map.get(r['车牌号'])
+    if old_r and old_r.get('是否达标') == '达标' and r.get('是否达标') != '达标':
+        flip.append(r)
+
+groups = defaultdict(lambda: defaultdict(int))
+for r in flip:
+    groups[r.get('考核目标') or '未知'][(r.get('部门名称') or '', r.get('销售经理') or '')] += 1
+
+wb = openpyxl.load_workbook(JUL)
+ref = openpyxl.load_workbook(JUN)
+ws = wb['测试与考核情况说明']
+rws = ref['测试与考核情况说明']
+
+# 清空原说明页后，按6月视觉规范重建。
+for merged in list(ws.merged_cells.ranges):
+    ws.unmerge_cells(str(merged))
+ws.delete_rows(1, ws.max_row)
+for col in range(1, 8):
+    letter = openpyxl.utils.get_column_letter(col)
+    ws.column_dimensions[letter].width = rws.column_dimensions[letter].width
+
+total_plates = len({r['车牌号'] for r in new})
+interv_plates = len({r['车牌号'] for r in intervention})
+total_tbox = sum(float(r.get('原TBOX里程(km)') or 0) for r in new)
+total_actual = sum(float(r.get('实际行驶里程(km)') or 0) for r in new)
+total_test = sum(float(r.get('测试里程(km)') or 0) for r in new)
+interv_tbox = sum(float(r.get('原TBOX里程(km)') or 0) for r in intervention)
+interv_actual = sum(float(r.get('实际行驶里程(km)') or 0) for r in intervention)
+interv_test = sum(float(r.get('测试里程(km)') or 0) for r in intervention)
+
+# 标题及说明：深蓝标题、浅黄说明，与6月一致。
+ws.merge_cells('A1:G1'); ws['A1'] = '2026年7月 业务考核 — 测试与考核情况说明'; clone_style(rws['A1'], ws['A1'])
+ws.row_dimensions[1].height = rws.row_dimensions[1].height
+ws.merge_cells('A2:G3'); ws['A2'] = ('7月业务考核车辆中，测试车辆统一以指定GPS日里程按考核期汇总替换TBOX里程，差额作为测试里程剥离且最低取0。\n'
+    '粤AG37316因GPS缺失，经业务确认暂按0核算。'); clone_style(rws['A2'], ws['A2'])
+ws.row_dimensions[2].height = rws.row_dimensions[2].height
+
+ws.merge_cells('A5:G5'); ws['A5'] = '一、7月业务考核与干预情况'; clone_style(rws['A5'], ws['A5'])
+headers = ['项目', '7月业务考核车辆', '涉及测试的车辆', '占比']
+for c, v in enumerate(headers, 1): ws.cell(6,c,v); clone_style(rws.cell(6,c), ws.cell(6,c))
+stats = [
+ ('车辆数(辆)', total_plates, interv_plates),
+ ('产生考核记录数(条)', len(new), len(intervention)),
+ ('仪表盘里程(km)', round(total_tbox), round(interv_tbox)),
+ ('业务考核里程 GPS(km)', round(total_actual), round(interv_actual)),
+ ('测试里程(km)', round(total_test), round(interv_test)),
+]
+for rr, (label, allv, partv) in enumerate(stats, 7):
+    for c, v in enumerate([label, allv, partv, partv/allv if allv else 0], 1):
+        ws.cell(rr,c,v); clone_style(rws.cell(7 + ((rr-7)%2),c), ws.cell(rr,c))
+
+# “测试里程”比例改为测试车辆剔除率，避免原100%被理解成车辆全部里程均为测试。
+ws['D6'] = '占比/剔除率'
+ws['D11'] = interv_test / interv_tbox if interv_tbox else 0
+negative_floor_adjustment = sum(max(0, float(r.get('实际行驶里程(km)') or 0) - float(r.get('原TBOX里程(km)') or 0)) for r in intervention)
+ws.merge_cells('A12:G12')
+ws['A12'] = (f'勾稽：测试里程 {round(interv_tbox):,} − {round(interv_actual):,} ＋ '
+             f'{round(negative_floor_adjustment):,}（负差归0调整）＝ {round(interv_test):,} km')
+clone_style(rws['A2'], ws['A12'])
+ws.row_dimensions[12].height = 22
+
+ws.merge_cells('A13:G13'); ws['A13']='二、干预车辆考核达标情况说明'; clone_style(rws['A13'],ws['A13'])
+ws.merge_cells('A14:G14'); ws['A14']=f'共 {len(intervention)} 条考核记录涉及里程干预；替换为GPS里程后，共 {len(flip)} 条考核车辆记录由“达标”转判为“未达标”。按车型分类如下：'; clone_style(rws['A14'],ws['A14'])
+
+target_order = ['羚牛136辆4.5T冷链车','羚牛100辆18T','交投40辆4.5T普货','交投190辆4.5T冷链车','恒运50辆4.5T普货','恒运50辆4.5T 普货']
+rn = 16
+for target in target_order:
+    items = sorted(groups.get(target, {}).items())
+    if not items: continue
+    total = sum(v for _, v in items)
+    ws.merge_cells(start_row=rn,start_column=1,end_row=rn,end_column=3)
+    ws.cell(rn,1,f'◎{target}（{total} 辆）'); clone_style(rws['A15'],ws.cell(rn,1)); rn += 1
+    for c,v in enumerate(['部门','销售经理','数量'],1): ws.cell(rn,c,v); clone_style(rws.cell(16,c),ws.cell(rn,c))
+    rn += 1
+    for idx, ((dept,sales),cnt) in enumerate(items):
+        for c,v in enumerate([dept,sales,cnt],1): ws.cell(rn,c,v); clone_style(rws.cell(17 + idx%2,c),ws.cell(rn,c))
+        rn += 1
+    rn += 1
+
+# 数据来源说明沿用6月蓝色区域。
+ws.merge_cells(start_row=rn,start_column=1,end_row=rn,end_column=7); ws.cell(rn,1,'三、里程三列数据来源说明'); clone_style(rws['A13'],ws.cell(rn,1)); rn += 1
+for c,v in enumerate(['列','字段名','数据来源','说明'],1): ws.cell(rn,c,v); clone_style(rws.cell(6,c),ws.cell(rn,c))
+rn += 1
+descs = [
+ ('O','原TBOX里程(km)','车辆 T-Box / 仪表盘读数','业务系统原始TBOX里程，包含运营、测试及调试里程。'),
+ ('P','实际行驶里程(km)','指定GPS日里程文件','按考核日期逐日汇总GPS里程；粤AG37316经确认暂按0核算。'),
+ ('Q','测试里程(km)','O − P（最低取0）','原TBOX里程与GPS里程的非负差额，不计入考核。'),
+]
+for idx,row in enumerate(descs):
+    for c,v in enumerate(row,1): ws.cell(rn,c,v); clone_style(rws.cell(28 + idx,c),ws.cell(rn,c))
+    ws.row_dimensions[rn].height = rws.row_dimensions[28 + idx].height
+    rn += 1
+
+ws.sheet_view.showGridLines = rws.sheet_view.showGridLines
+ws.freeze_panes = rws.freeze_panes
+wb.save(JUL)
+wb.close(); ref.close()
+print(f'updated={JUL}')
+print(f'rows={rn-1} flip={len(flip)} groups={sum(len(v) for v in groups.values())}')
diff --git a/full_pipeline_jun.py b/full_pipeline_jun.py
new file mode 100644
index 0000000..3701629
--- /dev/null
+++ b/full_pipeline_jun.py
@@ -0,0 +1,793 @@
+#!/usr/bin/env python3
+"""
+全流程：用GPS汇总文件替换测试车辆里程 → 重新核算6月 → 离职转嫁 → 输出
+输出目录: 20260713_1500/
+"""
+import openpyxl, os, shutil, xlrd
+from openpyxl.styles import Font, PatternFill, Alignment, Border, Side
+from openpyxl.utils import get_column_letter
+from collections import defaultdict
+from datetime import datetime, timedelta
+from calc_engine import *
+import excel_writer as ew
+
+os.chdir('/Users/kkfluous/Downloads')
+GPS_BASE = 'gps里程汇总2026年1-6月 3'
+OUT_DIR = '20260713_1500'
+os.makedirs(OUT_DIR, exist_ok=True)
+
+# ============================================================
+# Step 1: Build GPS daily mileage lookup from all 4 providers
+# ============================================================
+print("=" * 60)
+print("Step 1: 构建GPS日里程映射")
+print("=" * 60)
+
+# {plate: {day_number (1-30): mileage_km}}
+gps_daily = defaultdict(lambda: defaultdict(float))
+
+# --- G7 ---
+print("\n读取 G7...")
+fp_g7 = os.path.join(GPS_BASE, 'G7/202606_G7.xlsx')
+wb = openpyxl.load_workbook(fp_g7, data_only=True)
+ws = wb.active
+h_g7 = [c.value for c in next(ws.iter_rows(min_row=1, max_row=1))]
+# Find daily columns: 06月01日 ... 06月30日
+g7_day_cols = {}
+for i, h in enumerate(h_g7):
+    if h and '月' in str(h) and '日' in str(h):
+        # Parse "06月15日" -> day=15
+        import re
+        m = re.search(r'(\d+)月(\d+)日', str(h))
+        if m:
+            day = int(m.group(2))
+            g7_day_cols[day] = i
+
+print(f"  G7日期列: {len(g7_day_cols)} 天, 共 {ws.max_row-1} 辆车")
+g7_count = 0
+for row in ws.iter_rows(min_row=2, values_only=True):
+    plate = str(row[0]).strip() if row[0] else ''
+    if not plate: continue
+    for day, col in g7_day_cols.items():
+        val = float(row[col] or 0)
+        if val > 0:
+            gps_daily[plate][day] += val
+    g7_count += 1
+wb.close()
+print(f"  G7: {g7_count} 辆车")
+
+# --- 东方北斗 ---
+print("\n读取 东方北斗...")
+fp_df = os.path.join(GPS_BASE, '东方北斗/202606_东方北斗.xlsx')
+wb = openpyxl.load_workbook(fp_df, data_only=True)
+ws = wb.active
+h_df = [c.value for c in next(ws.iter_rows(min_row=3, max_row=3))]
+# Daily columns: 1号(Col 11) ... 30号(Col 40)
+df_day_cols = {}
+for i, h in enumerate(h_df):
+    if h and '号' in str(h) and str(h)[0].isdigit():
+        try:
+            day = int(str(h).replace('号', ''))
+            df_day_cols[day] = i
+        except: pass
+
+print(f"  东方北斗日期列: {len(df_day_cols)} 天")
+df_count = 0
+for row in ws.iter_rows(min_row=4, values_only=True):
+    plate = str(row[5]).strip() if row[5] else ''
+    if not plate: continue
+    for day, col in df_day_cols.items():
+        val = float(row[col] or 0)
+        if val > 0:
+            gps_daily[plate][day] += val
+    df_count += 1
+wb.close()
+print(f"  东方北斗: {df_count} 辆车")
+
+# --- 信达GPS (daily detail format) ---
+print("\n读取 信达GPS...")
+fp_xd = os.path.join(GPS_BASE, '信达GPS/202606_信达.xlsx')
+wb = openpyxl.load_workbook(fp_xd, data_only=True)
+ws = wb.active
+xd_count = 0
+for row in ws.iter_rows(min_row=4, values_only=True):
+    plate = str(row[7]).strip() if row[7] else ''
+    date_str = str(row[14]).strip() if row[14] else ''
+    mileage = float(row[15] or 0)
+    if not plate or not date_str or mileage <= 0: continue
+    try:
+        # Parse date: "2026-06-01"
+        dt = datetime.strptime(date_str[:10], '%Y-%m-%d')
+        day = dt.day
+        gps_daily[plate][day] += mileage
+        xd_count += 1
+    except: pass
+wb.close()
+print(f"  信达GPS: {xd_count} 条日记录")
+
+# --- 广安北斗 (monthly only - xls) ---
+print("\n读取 广安北斗...")
+fp_ga = os.path.join(GPS_BASE, '广安北斗/202606_广安北斗.xls')
+gdwb = xlrd.open_workbook(fp_ga)
+gdws = gdwb.sheet_by_index(0)
+ga_count = 0
+for r in range(7, gdws.nrows):
+    plate = str(gdws.cell_value(r, 3)).strip()
+    mileage = float(gdws.cell_value(r, 11) or 0)
+    if not plate or mileage <= 0: continue
+    # Spread across all 30 days proportionally? No - store as a special total
+    # Better: store at day 0 as "full month total"
+    if 0 not in gps_daily[plate] or gps_daily[plate][0] < mileage:
+        gps_daily[plate][0] = mileage  # day 0 = full month total marker
+    ga_count += 1
+print(f"  广安北斗: {ga_count} 辆车 (月度汇总)")
+
+total_plates = len(gps_daily)
+print(f"\nGPS映射总计: {total_plates} 辆车")
+
+# ============================================================
+# Step 2: Read test vehicle plates
+# ============================================================
+print("\n" + "=" * 60)
+print("Step 2: 提取测试车辆清单")
+print("=" * 60)
+
+# Extract from干预情况说明
+fp_interv = '租赁任务考核_2026年6月_干预情况说明.xlsx'
+if os.path.exists(fp_interv):
+    wb = openpyxl.load_workbook(fp_interv, data_only=True)
+    ws = wb['涉及测试的业务考核车辆']
+    h = [c.value for c in next(ws.iter_rows(min_row=1, max_row=1))]
+    plate_col = h.index('车牌号')
+    test_plates = set()
+    for row in ws.iter_rows(min_row=2, values_only=True):
+        p = row[plate_col]
+        if p: test_plates.add(str(p).strip())
+    wb.close()
+else:
+    # Fallback: vehicles with test mileage > 0 in current source
+    wb = openpyxl.load_workbook('租赁任务考核_2026年6月.xlsx', data_only=True)
+    ws = wb['业务考核视图']
+    h = [c.value for c in next(ws.iter_rows(min_row=1, max_row=1))]
+    pcol = h.index('车牌号')
+    tcol = h.index('测试里程(km)')
+    test_plates = set()
+    for row in ws.iter_rows(min_row=2, values_only=True):
+        if float(row[tcol] or 0) > 0:
+            test_plates.add(str(row[pcol]).strip())
+    wb.close()
+
+print(f"测试车辆: {len(test_plates)} 辆")
+
+# Check GPS coverage
+gps_hit = sum(1 for p in test_plates if p in gps_daily)
+gps_miss = test_plates - set(gps_daily.keys())
+print(f"GPS命中: {gps_hit}, GPS缺失: {len(gps_miss)}")
+if gps_miss:
+    print(f"缺失车辆(前10): {sorted(gps_miss)[:10]}")
+
+# ============================================================
+# Step 3: Replace GPS mileage for test vehicles
+# ============================================================
+print("\n" + "=" * 60)
+print("Step 3: 替换测试车辆实际行驶里程")
+print("=" * 60)
+
+# Start from current source file (has correct structure with TBOX/GPS/测试 columns)
+src_fp = '租赁任务考核_2026年6月.xlsx'
+wb_src = openpyxl.load_workbook(src_fp)
+ws_src = wb_src['业务考核视图']
+h_src = [c.value for c in next(ws_src.iter_rows(min_row=1, max_row=1))]
+
+# Column indices
+col_plate = h_src.index('车牌号')
+col_tbox = h_src.index('原TBOX里程(km)')
+col_actual = h_src.index('实际行驶里程(km)')
+col_test = h_src.index('测试里程(km)')
+col_rate = h_src.index('完成率(%)')
+col_dabiao = h_src.index('是否达标')
+col_start = h_src.index('考核开始日期')
+col_end = h_src.index('考核结束日期')
+col_target_km = h_src.index('应考核里程(km)')
+
+YELLOW_FILL = PatternFill(start_color='FFFF00', end_color='FFFF00', fill_type='solid')
+
+updated_count = 0
+no_gps_count = 0
+records = []
+
+for row_idx, row in enumerate(ws_src.iter_rows(min_row=2), start=2):
+    plate = str(row[col_plate].value or '').strip()
+    tbox_val = float(row[col_tbox].value or 0)
+    start_date = row[col_start].value
+    end_date = row[col_end].value
+    target_km = float(row[col_target_km].value or 0)
+
+    if plate not in test_plates:
+        records.append((row_idx, plate, False, None, None))
+        continue
+
+    # Calculate GPS mileage within考核日期区间
+    gps_sum = 0.0
+    gps_data = gps_daily.get(plate, {})
+
+    if 0 in gps_data and len(gps_data) == 1:
+        # 广安北斗: only has full month total
+        gps_sum = gps_data[0]
+    elif start_date and end_date:
+        # Parse dates and sum daily GPS
+        if isinstance(start_date, datetime) and isinstance(end_date, datetime):
+            start_day = start_date.day
+            end_day = end_date.day
+            for day in range(start_day, end_day + 1):
+                gps_sum += gps_data.get(day, 0)
+        elif isinstance(start_date, str):
+            try:
+                sd = datetime.strptime(str(start_date)[:10], '%Y-%m-%d')
+                ed = datetime.strptime(str(end_date)[:10], '%Y-%m-%d')
+                for day in range(sd.day, ed.day + 1):
+                    gps_sum += gps_data.get(day, 0)
+            except: pass
+        else:
+            # No dates, use full month
+            gps_sum = sum(gps_data.values())
+    else:
+        gps_sum = sum(gps_data.values())
+
+    if gps_sum <= 0 and tbox_val <= 0:
+        no_gps_count += 1
+        records.append((row_idx, plate, True, tbox_val, tbox_val))
+        continue
+
+    # Test mileage = max(0, TBOX - GPS)
+    test_mileage = max(0, tbox_val - gps_sum)
+
+    # Completion rate
+    rate = (gps_sum / target_km * 100) if target_km > 0 else 0
+
+    # 达标判断
+    dabiao = '达标' if gps_sum >= target_km and target_km > 0 else '未达标'
+
+    # Update cells
+    row[col_actual].value = round(gps_sum, 2)
+    row[col_test].value = round(test_mileage, 2)
+    row[col_rate].value = round(rate, 2)
+    row[col_dabiao].value = dabiao
+
+    # Yellow highlight
+    for cell in row:
+        cell.fill = YELLOW_FILL
+
+    updated_count += 1
+    records.append((row_idx, plate, True, gps_sum, test_mileage))
+
+print(f"测试车辆记录: {updated_count} 条已更新, {no_gps_count} 条无GPS数据")
+
+# Save updated source file
+new_src = os.path.join(OUT_DIR, '租赁任务考核_2026年6月.xlsx')
+wb_src.save(new_src)
+print(f"✅ 新数据源: {new_src}")
+
+# ============================================================
+# Step 4: Run 核算
+# ============================================================
+print("\n" + "=" * 60)
+print("Step 4: 运行核算")
+print("=" * 60)
+
+# Temporarily copy source to working location
+shutil.copy(new_src, '租赁任务考核_2026年6月.xlsx')
+
+# Re-run main.py logic
+print("读取源数据...")
+D = {}
+for m in range(1, 7):
+    fp = f'租赁任务考核_2026年{m}月.xlsx'
+    if os.path.exists(fp):
+        D[m] = read_file(fp, m)
+        print(f"  {m}月: {len(D[m])}条")
+
+G = {m: grp(D[m], m) for m in D}
+calc_jan_carryover(G[1])
+all_data = {}
+all_data[1] = {'当月': [{'车牌':k[0],'销售':g['销售'],'部门':g['部门'],'额':g['奖金']}
+                        for k,g in G[1].items() if g['奖金'] > 0]}
+
+for m in range(2, 7):
+    if m not in G: continue
+    data_m, G[m] = calc_month(m, G, all_data, G[m-1], G[m])
+
+vehicle_payments, vehicle_info = collect_vehicle_payments(G, all_data)
+master_vehicles = read_master_vehicles()
+
+# Loss data
+loss_data = {}
+for m in D:
+    ld = read_loss_data(m)
+    if ld:
+        loss_data[m] = ld
+    else:
+        from collections import defaultdict as dd
+        ld_tmp = dd(lambda: '否')
+        for r in D[m]:
+            c = r.get('客户名称')
+            if c: ld_tmp[c] = '否'
+        loss_data[m] = ld_tmp
+
+plate_client = get_vehicle_client_map(D)
+
+all_persons = {}
+for m in G:
+    for k, g in G[m].items():
+        all_persons[g['销售']] = g['部门']
+
+# Print summary
+settle_month = 6
+month_data = all_data[settle_month]
+total_6 = sum(sum(d['额'] for d in v) for v in month_data.values())
+print(f"\n6月考核应发: {total_6:.2f}")
+
+# Generate核算 file
+print(f"\n生成 {settle_month}月核算文件...")
+wb = openpyxl.Workbook()
+ew.write_rules_sheet(wb)
+ew.write_detail_sheet(wb, D[settle_month], settle_month)
+ew.write_calc_process_generic(wb, settle_month, G, all_data)
+ew.write_vehicle_tracking_sheet(wb, settle_month, G, master_vehicles, vehicle_payments, vehicle_info, loss_data[settle_month], plate_client)
+
+class _AllPass:
+    def get(self, _k, _d=None): return '否'
+    def __bool__(self): return True
+payment_records = ew.build_payment_records(settle_month, month_data, {m: _AllPass() for m in D}, plate_client)
+ew.write_payment_record_sheet(wb, settle_month, payment_records)
+ew.write_summary_from_records(wb, settle_month, payment_records)
+
+for person in sorted(all_persons.keys()):
+    ew.write_salesperson_sheet(wb, person, all_persons[person], settle_month, D, G, month_data, vehicle_payments)
+
+if 'Sheet' in wb.sheetnames:
+    del wb['Sheet']
+
+calc_fp = os.path.join(OUT_DIR, '里程任务考核_6月核算.xlsx')
+wb.save(calc_fp)
+print(f"✅ 核算文件: {calc_fp}")
+
+# ============================================================
+# Step 5: 离职转嫁 + 格式重建
+# ============================================================
+print("\n" + "=" * 60)
+print("Step 5: 离职转嫁 + 汇总格式重建")
+print("=" * 60)
+
+RESIGNED = {'赵连飞', '伍仲文', '岑彦'}
+TRANSFERS = [
+    ('刘念念', '赵连飞', 2260.00),
+    ('钟祥',   '伍仲文、岑彦', 180.00),
+]
+TRANSFER_TOTAL = sum(t[2] for t in TRANSFERS)
+TRANSFER_MAP = {'赵连飞': ('刘念念', 2260.00), '伍仲文': ('钟祥', 150.00), '岑彦': ('钟祥', 30.00)}
+
+# Styles
+HEAD_FILL = PatternFill('solid', fgColor='4472C4')
+HEAD_FONT = Font(bold=True, color='FFFFFF', size=10, name='宋体')
+SUB_FILL  = PatternFill('solid', fgColor='D9E1F2')
+TITLE_FNT = Font(bold=True, size=12, name='宋体')
+SECTION_FNT = Font(bold=True, size=11, name='宋体')
+NOTE_FNT  = Font(size=10, color='C00000', name='宋体')
+NORM_FNT  = Font(size=10, name='宋体')
+BOLD_FNT  = Font(bold=True, size=10, name='宋体')
+THIN = Side('thin', color='B4B4B4')
+BORD = Border(THIN, THIN, THIN, THIN)
+CC = Alignment(horizontal='center', vertical='center', wrap_text=True)
+LL = Alignment(horizontal='left', vertical='center', wrap_text=True)
+
+def hdr_row(ws, r, headers):
+    for c, h in enumerate(headers, 1):
+        cell = ws.cell(row=r, column=c, value=h)
+        cell.font = HEAD_FONT; cell.fill = HEAD_FILL; cell.alignment = CC; cell.border = BORD
+
+def data_row(ws, r, values, bold=False):
+    for c, v in enumerate(values, 1):
+        cell = ws.cell(row=r, column=c, value=v)
+        cell.font = BOLD_FNT if bold else NORM_FNT; cell.alignment = CC; cell.border = BORD
+        if isinstance(v, float): cell.number_format = '#,##0.00'
+
+def sub_total(ws, r, values):
+    for c, v in enumerate(values, 1):
+        cell = ws.cell(row=r, column=c, value=v)
+        cell.font = BOLD_FNT; cell.fill = SUB_FILL; cell.alignment = CC; cell.border = BORD
+        if isinstance(v, float): cell.number_format = '#,##0.00'
+
+# Parse existing summary
+ws_old = wb['6月汇总']
+person_dept = {}
+for r in range(1, ws_old.max_row + 1):
+    v1 = str(ws_old.cell(row=r, column=1).value or '')
+    if '6月最终发放（按销售人员）' in v1:
+        for r2 in range(r+1, ws_old.max_row + 1):
+            name = str(ws_old.cell(row=r2, column=1).value or '')
+            dept = str(ws_old.cell(row=r2, column=2).value or '')
+            if name == '合计': break
+            if name and name != '销售人员' and dept:
+                person_dept[name] = dept
+        break
+
+# Parse考核应发 subsections
+subsections = []
+current_title = None; current_rows = []
+for r in range(1, ws_old.max_row + 1):
+    v1 = str(ws_old.cell(row=r, column=1).value or '')
+    if v1.startswith('考核应发-'):
+        if current_title and current_rows:
+            subsections.append((current_title, current_rows))
+        current_title = v1; current_rows = []
+        continue
+    if current_title:
+        name = str(ws_old.cell(row=r, column=1).value or '')
+        vehicles = ws_old.cell(row=r, column=2).value
+        amount = ws_old.cell(row=r, column=3).value
+        if name == '总计':
+            subsections.append((current_title, current_rows))
+            current_title = None; current_rows = []
+        elif name and name != '销售人员':
+            current_rows.append((name, vehicles, amount))
+
+# Filter out resigned
+filtered_subsections = []
+grand_total = 0.0
+for title, rows in subsections:
+    filtered = [(n, v, a) for (n, v, a) in rows if n not in RESIGNED]
+    new_veh = sum(int(v or 0) for _, v, _ in filtered)
+    new_amt = sum(float(a or 0) for _, _, a in filtered)
+    grand_total += new_amt
+    filtered_subsections.append((title, filtered, new_veh, new_amt))
+
+# Aggregate最终发放
+person_data = defaultdict(lambda: {'vehicles': 0, 'amount': 0.0})
+for title, rows, _, _ in filtered_subsections:
+    for name, vehicles, amount in rows:
+        person_data[name]['vehicles'] += int(vehicles or 0)
+        person_data[name]['amount'] += float(amount or 0)
+
+for resigned, (receiver, amt) in TRANSFER_MAP.items():
+    person_data[receiver]['amount'] += amt
+
+for name in RESIGNED:
+    person_data.pop(name, None)
+
+final_total = sum(d['amount'] for d in person_data.values())
+
+dept_order = ['业务二部', '业务三部', '业务五部', '业务六部']
+def sort_key(item):
+    name = item[0]
+    dept = person_dept.get(name, '')
+    try: di = dept_order.index(dept)
+    except: di = 99
+    return (di, name)
+
+sorted_persons = sorted(person_data.items(), key=sort_key)
+
+dept_totals = defaultdict(float)
+for name, data in sorted_persons:
+    dept = person_dept.get(name, '')
+    dept_totals[dept] += data['amount']
+
+# Rebuild 6月汇总 sheet
+ws = wb['6月汇总']
+ws.delete_rows(1, ws.max_row)
+rn = 1
+
+# 一、考核应发明细
+ws.merge_cells(start_row=rn, start_column=1, end_row=rn, end_column=3)
+ws.cell(row=rn, column=1, value='一、考核应发明细').font = TITLE_FNT
+rn += 2
+
+for title, rows, total_veh, total_amt in filtered_subsections:
+    ws.cell(row=rn, column=1, value=title).font = SECTION_FNT
+    rn += 1
+    hdr_row(ws, rn, ['销售人员', '车辆数', '金额'])
+    rn += 1
+    sorted_rows = sorted(rows, key=lambda x: sort_key((x[0],)) if x[0] in person_dept else (99, x[0]))
+    for name, vehicles, amount in sorted_rows:
+        data_row(ws, rn, [name, vehicles, round(float(amount or 0), 2)])
+        rn += 1
+    sub_total(ws, rn, ['总计', total_veh, round(total_amt, 2)])
+    rn += 2
+
+ws.merge_cells(start_row=rn, start_column=1, end_row=rn, end_column=2)
+ws.cell(row=rn, column=1, value='考核应发合计').font = BOLD_FNT
+ws.cell(row=rn, column=1).alignment = CC
+ws.cell(row=rn, column=3, value=round(grand_total, 2)).font = BOLD_FNT
+ws.cell(row=rn, column=3).alignment = CC; ws.cell(row=rn, column=3).number_format = '#,##0.00'
+for c in range(1, 4):
+    ws.cell(row=rn, column=c).border = BORD; ws.cell(row=rn, column=c).fill = SUB_FILL
+rn += 2
+
+# 二、离职人员转嫁收入
+ws.merge_cells(start_row=rn, start_column=1, end_row=rn, end_column=3)
+ws.cell(row=rn, column=1, value='二、离职人员转嫁收入').font = TITLE_FNT
+rn += 1
+hdr_row(ws, rn, ['接收人', '来源', '转嫁金额'])
+rn += 1
+for receiver, source, amt in TRANSFERS:
+    data_row(ws, rn, [receiver, source, amt]); rn += 1
+sub_total(ws, rn, ['合计', '', TRANSFER_TOTAL]); rn += 2
+
+# 三、最终发放
+ws.merge_cells(start_row=rn, start_column=1, end_row=rn, end_column=3)
+ws.cell(row=rn, column=1, value='三、最终发放').font = TITLE_FNT
+rn += 2
+
+ws.cell(row=rn, column=1, value='最终发放明细').font = SECTION_FNT; rn += 1
+hdr_row(ws, rn, ['销售人员', '车辆数', '金额']); rn += 1
+for name, data in sorted_persons:
+    data_row(ws, rn, [name, data['vehicles'], round(data['amount'], 2)]); rn += 1
+sub_total(ws, rn, ['总计', sum(d['vehicles'] for _, d in sorted_persons), round(final_total, 2)]); rn += 2
+
+ws.cell(row=rn, column=1, value='6月最终发放（按销售人员）').font = SECTION_FNT; rn += 1
+hdr_row(ws, rn, ['销售人员', '部门名称', '最终发放']); rn += 1
+for name, data in sorted_persons:
+    dept = person_dept.get(name, '')
+    data_row(ws, rn, [name, dept, round(data['amount'], 2)]); rn += 1
+sub_total(ws, rn, ['合计', '', round(final_total, 2)]); rn += 2
+
+ws.cell(row=rn, column=1, value='6月最终发放（按部门）').font = SECTION_FNT; rn += 1
+hdr_row(ws, rn, ['部门名称', '最终发放', '']); rn += 1
+for dept in dept_order:
+    if dept in dept_totals:
+        data_row(ws, rn, [dept, round(dept_totals[dept], 2), '']); rn += 1
+sub_total(ws, rn, ['合计', round(sum(dept_totals.values()), 2), '']); rn += 2
+
+# 总览
+ws.cell(row=rn, column=1, value='总览').font = SECTION_FNT; rn += 1
+data_row(ws, rn, ['考核应发', round(final_total, 2), '']); rn += 1
+data_row(ws, rn, ['亏损拦截', 0, '']); rn += 1
+data_row(ws, rn, ['最终发放', round(final_total, 2), '']); rn += 2
+
+# 备注
+note = '备注：赵连飞（2026-04-30离职）、伍仲文（2026-05-09离职）、岑彦（2026-05-09离职）\n三人已从上述汇总中剔除，其6月当月达标奖金已转嫁至新销售；其余结转、补发、累计补发等一律不发了。'
+ws.merge_cells(start_row=rn, start_column=1, end_row=rn, end_column=3)
+ws.cell(row=rn, column=1, value=note).font = NOTE_FNT
+ws.cell(row=rn, column=1).alignment = Alignment(horizontal='left', vertical='center', wrap_text=True)
+ws.row_dimensions[rn].height = 36
+
+# Column widths
+ws.column_dimensions['A'].width = 16
+ws.column_dimensions['B'].width = 18
+ws.column_dimensions['C'].width = 16
+
+# Rename resigned sheets
+for old, new in [('二部-赵连飞','【已离职】二部-赵连飞'),('六部-伍仲文','【已离职】六部-伍仲文'),('六部-岑彦','【已离职】六部-岑彦')]:
+    if old in wb.sheetnames:
+        wb[old].title = new
+        wb.move_sheet(new, offset=len(wb.sheetnames)-1-wb.sheetnames.index(new))
+
+wb.save(calc_fp)
+print(f"✅ 核算文件(含转嫁): {calc_fp}")
+
+# ============================================================
+# Step 6: 干预情况说明 (copy from existing)
+# ============================================================
+print("\n" + "=" * 60)
+print("Step 6: 干预情况说明")
+print("=" * 60)
+
+old_interv = '租赁任务考核_2026年6月_干预情况说明.xlsx'
+if os.path.exists(old_interv):
+    shutil.copy(old_interv, os.path.join(OUT_DIR, '租赁任务考核_2026年6月_干预情况说明.xlsx'))
+    print(f"✅ 干预说明已复制")
+
+# ============================================================
+# Step 7: 离职明细 (regenerate with new amounts)
+# ============================================================
+print("\n" + "=" * 60)
+print("Step 7: 离职人员处理明细")
+print("=" * 60)
+
+# Collect resigned data from all_data[6]
+detail_records = []
+for ptype, items in month_data.items():
+    for it in items:
+        if it['销售'] not in RESIGNED: continue
+        name = it['销售']
+        amount = it['额']
+        is_dangyue = (ptype == '当月')
+        transfer_to = TRANSFER_MAP[name][0]
+        handling = f'转{transfer_to}(需确认)' if is_dangyue else '不发了'
+        transfer_disp = transfer_to if is_dangyue else '—'
+        note_txt = '6月当月达标，转新销售' if is_dangyue else f'{ptype}，不发了'
+
+        client = ''; target_name = ''
+        for r in D.get(6, []):
+            if r['车牌号'] == it['车牌'] and r['销售经理'] == name:
+                client = r.get('客户名称', ''); target_name = r.get('考核目标', '')
+                break
+
+        g6 = G[6].get((it['车牌'], name))
+        monthly_bonus = g6['奖励达标'] if g6 else 0
+        days = g6['天数'] if g6 else ''
+
+        if ptype in ('当月', f'累计补发{settle_month}月', '结转'):
+            loss_month = settle_month
+        elif ptype.startswith('补发'):
+            try: loss_month = int(ptype.replace('补发','').replace('月',''))
+            except: loss_month = settle_month
+        else: loss_month = settle_month
+
+        detail_records.append({
+            'name': name, 'plate': it['车牌'], 'pay_type': ptype,
+            'amount': amount, 'handling': handling, 'transfer': transfer_disp,
+            'note': note_txt, 'client': client, 'loss_month': loss_month,
+            'loss_status': '否', 'monthly_bonus': monthly_bonus,
+            'days': days, 'target': target_name,
+        })
+
+print(f"离职人员发放记录: {len(detail_records)} 条")
+
+# Generate resignation workbook (simplified)
+wb_res = openpyxl.Workbook()
+
+# Sheet 1: 总览
+ws1 = wb_res.active; ws1.title = '离职人员发放处理总览'
+ws1.merge_cells('A1:G1')
+ws1.cell(row=1, column=1, value='2026年6月核算 — 离职人员奖金处理方案').font = Font(bold=True, size=14, name='宋体')
+ws1.cell(row=1, column=1).alignment = CC
+ws1.row_dimensions[1].height = 30
+
+ws1.merge_cells('A2:G2')
+ws1.cell(row=2, column=1, value='处理规则：仅6月当月达标奖金转给新接手的销售（需与其他部门确认是否发放）；其余结转、补发、累计补发等一律不发了。').font = NOTE_FNT
+ws1.cell(row=2, column=1).alignment = LL
+
+r = 4
+for c, h in enumerate(['离职人员', '部门', '业务转嫁至', '不发了', '转新销售(需确认)', '原考核应发合计'], 1):
+    ws1.cell(row=r, column=c, value=h)
+    ws1.cell(row=r, column=c).font = HEAD_FONT; ws1.cell(row=r, column=c).fill = HEAD_FILL
+    ws1.cell(row=r, column=c).alignment = CC; ws1.cell(row=r, column=c).border = BORD
+r += 1
+
+dept_map = {'赵连飞': '业务二部', '伍仲文': '业务六部', '岑彦': '业务六部'}
+total_not = 0; total_trf = 0
+for name in ['赵连飞', '伍仲文', '岑彦']:
+    recs = [p for p in detail_records if p['name'] == name]
+    not_pay = sum(p['amount'] for p in recs if p['pay_type'] != '当月')
+    transfer = sum(p['amount'] for p in recs if p['pay_type'] == '当月')
+    total_not += not_pay; total_trf += transfer
+    data_row(ws1, r, [name, dept_map[name], TRANSFER_MAP[name][0], round(not_pay,2), round(transfer,2), round(not_pay+transfer,2)])
+    r += 1
+
+sub_total(ws1, r, ['合计', '', '', round(total_not,2), round(total_trf,2), round(total_not+total_trf,2)])
+r += 2
+
+ws1.merge_cells(start_row=r, start_column=1, end_row=r, end_column=6)
+ws1.cell(row=r, column=1, value='转嫁接收方汇总（需与其他部门确认是否发放）').font = Font(bold=True, size=12, color='1F4E78')
+ws1.cell(row=r, column=1).alignment = CC; r += 1
+
+for c, h in enumerate(['接收人', '来源离职人员', '转嫁金额', '备注'], 1):
+    ws1.cell(row=r, column=c, value=h)
+    ws1.cell(row=r, column=c).font = HEAD_FONT; ws1.cell(row=r, column=c).fill = HEAD_FILL
+    ws1.cell(row=r, column=c).alignment = CC; ws1.cell(row=r, column=c).border = BORD
+r += 1
+
+by_recv = defaultdict(lambda: {'sources':[], 'amt':0})
+for rec in detail_records:
+    if rec['pay_type'] == '当月':
+        t = rec['transfer']
+        by_recv[t]['sources'].append(rec['name'])
+        by_recv[t]['amt'] += rec['amount']
+
+for receiver in ['刘念念', '钟祥']:
+    d = by_recv[receiver]
+    data_row(ws1, r, [receiver, '、'.join(sorted(set(d['sources']))), round(d['amt'],2), '需与相关部门确认是否发放给新销售'])
+    r += 1
+
+for i, w in enumerate([14, 22, 18, 18, 18, 18], 1):
+    ws1.column_dimensions[get_column_letter(i)].width = w
+
+# Sheet 2: 逐条明细
+ws2 = wb_res.create_sheet('发放明细(逐条)')
+ws2.merge_cells('A1:M1')
+ws2.cell(row=1, column=1, value='离职人员 — 6月考核应发逐条明细').font = Font(bold=True, size=14, name='宋体')
+ws2.cell(row=1, column=1).alignment = CC
+
+dh = ['离职人员', '车牌号', '发放类型', '考核应发', '盈亏查询月', '客户名称', '客户盈亏', '处理方式', '转嫁接收人', '6月考核天数', '月度奖励', '考核目标', '备注']
+r2 = 3
+hdr_row(ws2, r2, dh); r2 += 1
+
+name_order = ['赵连飞', '伍仲文', '岑彦']
+detail_records.sort(key=lambda x: (name_order.index(x['name']), 0 if x['pay_type']=='当月' else 1, x['plate']))
+
+for i, rec in enumerate(detail_records):
+    vals = [rec['name'], rec['plate'], rec['pay_type'], round(rec['amount'],2), f"{rec['loss_month']}月",
+            rec['client'], rec['loss_status'], rec['handling'], rec['transfer'],
+            rec['days'] if rec['days'] else '', rec['monthly_bonus'] if rec['monthly_bonus'] else '', rec['target'], rec['note']]
+    data_row(ws2, r2, vals); ws2.row_dimensions[r2].height = 22; r2 += 1
+
+sub_total(ws2, r2, ['', '', '合计', round(sum(r['amount'] for r in detail_records),2), '', '', '', '', '', '', '', '', ''])
+
+for i, w in enumerate([10, 12, 14, 14, 10, 24, 8, 18, 14, 12, 10, 22, 28], 1):
+    ws2.column_dimensions[get_column_letter(i)].width = w
+
+# Personal sheets (3-5) + Transfer sheets (6-7) - simplified
+for name, title_txt, transfer_to in [
+    ('赵连飞', '赵连飞（2026-04-30离职，业务转→刘念念）', '刘念念'),
+    ('伍仲文', '伍仲文（2026-05-09离职，业务转→钟祥）', '钟祥'),
+    ('岑彦', '岑彦（2026-05-09离职，业务转→钟祥）', '钟祥'),
+]:
+    ws = wb_res.create_sheet(name)
+    ws.merge_cells('A1:K1')
+    ws.cell(row=1, column=1, value=title_txt).font = Font(bold=True, size=14, name='宋体')
+    ws.cell(row=1, column=1).alignment = CC
+
+    ph = ['车牌号', '发放类型', '考核应发', '盈亏查询月', '客户名称', '客户盈亏', '处理方式', '6月考核天数', '月度奖励', '考核目标', '备注']
+    r3 = 3
+    hdr_row(ws, r3, ph); r3 += 1
+
+    person_recs = [p for p in detail_records if p['name'] == name]
+    for rec in person_recs:
+        vals = [rec['plate'], rec['pay_type'], round(rec['amount'],2), f"{rec['loss_month']}月",
+                rec['client'], rec['loss_status'], rec['handling'],
+                rec['days'] if rec['days'] else '', rec['monthly_bonus'] if rec['monthly_bonus'] else '',
+                rec['target'], rec['note']]
+        data_row(ws, r3, vals); ws.row_dimensions[r3].height = 22; r3 += 1
+
+    dn = sum(p['amount'] for p in person_recs if p['pay_type']=='当月')
+    ot = sum(p['amount'] for p in person_recs if p['pay_type']!='当月')
+    sub_total(ws, r3, ['', '不发了小计', round(ot,2), '', '', '', '', '', '', '', '']); r3 += 1
+    sub_total(ws, r3, ['', f'转{transfer_to}小计', round(dn,2), '', '', '', '', '', '', '', '需与其他部门确认']); r3 += 1
+    sub_total(ws, r3, ['', '合计', round(dn+ot,2), '', '', '', '', '', '', '', ''])
+
+    for i, w in enumerate([12, 14, 14, 10, 26, 8, 18, 12, 10, 22, 30], 1):
+        ws.column_dimensions[get_column_letter(i)].width = w
+
+# Transfer detail sheets
+for title_txt, sources, receiver in [
+    ('转嫁至刘念念的奖金 — 来源：赵连飞(2026-04-30离职)', ['赵连飞'], '刘念念'),
+    ('转嫁至钟祥的奖金 — 来源：伍仲文(2026-05-09离职)、岑彦(2026-05-09离职)', ['伍仲文', '岑彦'], '钟祥'),
+]:
+    ws = wb_res.create_sheet(f'转{receiver}明细')
+    ws.merge_cells('A1:K1')
+    ws.cell(row=1, column=1, value=title_txt).font = Font(bold=True, size=14, name='宋体')
+    ws.cell(row=1, column=1).alignment = CC
+
+    ws.merge_cells('A2:K2')
+    ws.cell(row=2, column=1, value='⚠ 以下金额需与其他部门确认是否可发放给新销售').font = NOTE_FNT
+    ws.cell(row=2, column=1).alignment = LL
+
+    th = ['来源离职人员', '车牌号', '发放类型', '考核应发', '盈亏查询月', '客户名称', '客户盈亏', '6月考核天数', '月度奖励', '考核目标', '备注']
+    r4 = 4
+    hdr_row(ws, r4, th); r4 += 1
+
+    trf_recs = [p for p in detail_records if p['name'] in sources and p['pay_type']=='当月']
+    for rec in trf_recs:
+        vals = [rec['name'], rec['plate'], '当月', round(rec['amount'],2), f"{rec['loss_month']}月",
+                rec['client'], rec['loss_status'], rec['days'] if rec['days'] else '',
+                rec['monthly_bonus'] if rec['monthly_bonus'] else '', rec['target'], '需确认是否发给新销售']
+        data_row(ws, r4, vals); r4 += 1
+
+    sub_total(ws, r4, ['', '', '合计', round(sum(r['amount'] for r in trf_recs),2), '', '', '', '', '', '', ''])
+
+    for i, w in enumerate([14, 12, 14, 14, 10, 26, 8, 12, 10, 22, 22], 1):
+        ws.column_dimensions[get_column_letter(i)].width = w
+
+resign_fp = os.path.join(OUT_DIR, '离职人员6月奖金处理明细.xlsx')
+wb_res.save(resign_fp)
+print(f"✅ 离职明细: {resign_fp}")
+
+# ============================================================
+# Step 8: Summary
+# ============================================================
+print("\n" + "=" * 60)
+print("Step 8: 输出汇总")
+print("=" * 60)
+
+files = os.listdir(OUT_DIR)
+for f in sorted(files):
+    fp = os.path.join(OUT_DIR, f)
+    size_kb = os.path.getsize(fp) / 1024
+    print(f"  {f} ({size_kb:.0f} KB)")
+
+print(f"\n{'=' * 60}")
+print(f"6月考核应发: {total_6:.2f}")
+print(f"最终发放: {final_total:.2f}")
+print(f"全部文件已输出到 {OUT_DIR}/")
+print(f"{'=' * 60}")
diff --git a/gen_intervention_jun.py b/gen_intervention_jun.py
new file mode 100644
index 0000000..c4f67a4
--- /dev/null
+++ b/gen_intervention_jun.py
@@ -0,0 +1,281 @@
+#!/usr/bin/env python3
+"""生成6月干预情况说明"""
+import openpyxl, os, datetime
+from openpyxl.styles import Font, Alignment, Border, Side, PatternFill
+from openpyxl.utils import get_column_letter
+from collections import defaultdict
+
+os.chdir('/Users/kkfluous/Downloads')
+
+# ========== Styles ==========
+hdr_blue_font = Font(name='宋体', size=11, bold=True, color='FFFFFF')
+hdr_blue_fill = PatternFill(start_color='4472C4', end_color='4472C4', fill_type='solid')
+hdr_green_font = Font(name='宋体', size=11, bold=True, color='FFFFFF')
+hdr_green_fill = PatternFill(start_color='548235', end_color='548235', fill_type='solid')
+hdr_align = Alignment(horizontal='center', vertical='center', wrap_text=True)
+data_font = Font(name='宋体', size=10)
+data_align = Alignment(horizontal='center', vertical='center')
+title_font = Font(name='宋体', size=14, bold=True)
+subtitle_font = Font(name='宋体', size=10, bold=True)
+yellow_fill = PatternFill(start_color='FFFF00', end_color='FFFF00', fill_type='solid')
+light_yellow_fill = PatternFill(start_color='FFF2CC', end_color='FFF2CC', fill_type='solid')
+light_gray_fill = PatternFill(start_color='F2F2F2', end_color='F2F2F2', fill_type='solid')
+thin_border = Border(left=Side('thin'), right=Side('thin'), top=Side('thin'), bottom=Side('thin'))
+
+def set_cell(ws, row, col, value, font=None, fill=None, align=None, border=None, num_fmt=None):
+    cell = ws.cell(row=row, column=col, value=value)
+    if font: cell.font = font
+    if fill: cell.fill = fill
+    if align: cell.alignment = align
+    if border: cell.border = border
+    if num_fmt: cell.number_format = num_fmt
+    return cell
+
+# ========== Read Data ==========
+src = openpyxl.load_workbook('租赁任务考核_2026年6月.xlsx', data_only=True)
+ws_src = src['业务考核视图']
+h_src = [c.value for c in next(ws_src.iter_rows(min_row=1, max_row=1))]
+all_recs = []
+for row in ws_src.iter_rows(min_row=2, values_only=True):
+    r = dict(zip(h_src, row))
+    if r['车牌号']:
+        all_recs.append(r)
+src.close()
+
+# Read test vehicle list
+fp_test = '/Users/kkfluous/Library/Containers/com.tencent.xinWeChat/Data/Documents/xwechat_files/wxid_1704407055112_3af8/temp/RWTemp/2026-07/65b41c10aa83cabb7a365b2b8437c1c7/6月测试车辆.xlsx'
+wb_test = openpyxl.load_workbook(fp_test, data_only=True)
+ws_test = wb_test.active
+test_rows = list(ws_test.iter_rows(min_row=1, values_only=True))
+test_plates = set()
+for row in test_rows[1:]:
+    for ci in [0, 3, 6]:
+        p = str(row[ci]).strip() if row[ci] else ''
+        if p and p != 'None':
+            test_plates.add(p)
+wb_test.close()
+
+# Identify干预 records (test vehicles in source)
+intervention_recs = [r for r in all_recs if r['车牌号'] in test_plates]
+intervention_plates = set(r['车牌号'] for r in intervention_recs)
+
+# Stats
+total_vehicles = len(set(r['车牌号'] for r in all_recs))
+total_records = len(all_recs)
+interv_vehicles = len(intervention_plates)
+interv_records = len(intervention_recs)
+total_actual = sum(float(r['实际行驶里程(km)'] or 0) for r in all_recs)
+total_tbox = sum(float(r['原TBOX里程(km)'] or 0) for r in all_recs)
+total_test = sum(float(r['测试里程(km)'] or 0) for r in all_recs)
+interv_actual = sum(float(r['实际行驶里程(km)'] or 0) for r in intervention_recs)
+interv_tbox = sum(float(r['原TBOX里程(km)'] or 0) for r in intervention_recs)
+interv_test = sum(float(r['测试里程(km)'] or 0) for r in intervention_recs)
+
+# Flip analysis:达→未达标 (using GPS backup for original)
+bak = openpyxl.load_workbook('租赁任务考核_2026年6月_备份.xlsx', data_only=True)
+ws_bak = bak['业务考核视图']
+h_bak = [c.value for c in next(ws_bak.iter_rows(min_row=1, max_row=1))]
+bak_recs = {}
+for row in ws_bak.iter_rows(min_row=2, values_only=True):
+    r = dict(zip(h_bak, row))
+    if r['车牌号']:
+        bak_recs[r['车牌号']] = r
+bak.close()
+
+# For intervention vehicles: compare old达→new未达标
+flip_recs = []  # records that changed from达 to未达标
+for r in intervention_recs:
+    plate = r['车牌号']
+    old_rec = bak_recs.get(plate)
+    if old_rec:
+        old_q = old_rec.get('是否达标') == '达标'
+        new_q = r['是否达标'] == '达标'
+        if old_q and not new_q:
+            flip_recs.append(r)
+
+# Group flip by考核目标+部门+销售
+flip_by_target = defaultdict(lambda: defaultdict(lambda: defaultdict(int)))
+for r in flip_recs:
+    target = r.get('考核目标', '未知')
+    dept = r.get('部门名称', '')
+    sales = r.get('销售经理', '')
+    flip_by_target[target][dept][sales] += 1
+
+# ========== Create Workbook ==========
+wb = openpyxl.Workbook()
+
+# ===== Sheet 1: 测试与考核情况说明 =====
+ws1 = wb.active
+ws1.title = '测试与考核情况说明'
+
+# Column widths
+for col, w in enumerate([4, 30, 18, 18, 18, 18, 18], 1):
+    ws1.column_dimensions[get_column_letter(col)].width = w
+
+rn = 1
+
+# Title
+set_cell(ws1, rn, 1, '2026年6月 业务考核 — 测试与考核情况说明', title_font)
+ws1.merge_cells(start_row=rn, start_column=1, end_row=rn, end_column=5); rn += 1
+
+# Subtitle
+set_cell(ws1, rn, 1, '6月业务考核车辆中，部分车辆当月存在测试行为（仪表盘里程 > GPS 实际行驶里程），其差额作为测试里程剥离，不参与考核',
+         Font(name='宋体', size=10), light_yellow_fill)
+ws1.merge_cells(start_row=rn, start_column=1, end_row=rn, end_column=5); rn += 2
+
+# Section 1: Stats
+set_cell(ws1, rn, 1, '一、6月业务考核与干预情况', subtitle_font); rn += 1
+
+# Header
+for ci, h in enumerate(['项目', '6月业务考核车辆', '涉及测试的车辆', '占比'], 1):
+    set_cell(ws1, rn, ci, h, hdr_blue_font, hdr_blue_fill, hdr_align, thin_border)
+rn += 1
+
+# Data rows
+stats = [
+    ('车辆数(辆)', total_vehicles, interv_vehicles),
+    ('产生考核记录数(条)', total_records, interv_records),
+    ('仪表盘里程(km)', round(total_tbox), round(interv_tbox)),
+    ('业务考核里程 GPS(km)', round(total_actual), round(interv_actual)),
+    ('测试里程(km)', round(total_test), round(interv_test)),
+]
+for label, all_v, interv_v in stats:
+    ratio = interv_v / all_v if all_v else 0
+    set_cell(ws1, rn, 1, label, data_font, None, data_align, thin_border)
+    set_cell(ws1, rn, 2, all_v, data_font, None, data_align, thin_border)
+    set_cell(ws1, rn, 3, interv_v, data_font, None, data_align, thin_border)
+    set_cell(ws1, rn, 4, ratio, data_font, None, data_align, thin_border, '0.00%')
+    rn += 1
+
+rn += 2
+
+# Section 2: Flip details
+set_cell(ws1, rn, 1, '二、干预车辆考核达标情况说明', subtitle_font); rn += 1
+flip_total = len(flip_recs)
+summary_text = f'共 {interv_records} 条考核记录涉及里程干预；替换为GPS里程后，共 {flip_total} 条考核车辆记录由"达标"转判为"未达标"。按车型分类如下'
+set_cell(ws1, rn, 1, summary_text, Font(name='宋体', size=10))
+ws1.merge_cells(start_row=rn, start_column=1, end_row=rn, end_column=5); rn += 2
+
+# By target
+target_order = ['羚牛136辆4.5T冷链车', '羚牛100辆18T', '交投40辆4.5T普货', '交投190辆4.5T冷链车', '恒运50辆4.5T普货']
+for target in target_order:
+    if target not in flip_by_target:
+        continue
+    dept_sales = flip_by_target[target]
+    total_cnt = sum(sum(sales.values()) for sales in dept_sales.values())
+    
+    set_cell(ws1, rn, 1, f'▎{target}（{total_cnt} 辆）', subtitle_font)
+    ws1.merge_cells(start_row=rn, start_column=1, end_row=rn, end_column=5); rn += 1
+    
+    # Header: dept | sales | count | dept | sales | count
+    set_cell(ws1, rn, 1, '部门', hdr_green_font, hdr_green_fill, hdr_align, thin_border)
+    set_cell(ws1, rn, 2, '销售经理', hdr_green_font, hdr_green_fill, hdr_align, thin_border)
+    set_cell(ws1, rn, 3, '数量', hdr_green_font, hdr_green_fill, hdr_align, thin_border)
+    set_cell(ws1, rn, 4, '部门', hdr_green_font, hdr_green_fill, hdr_align, thin_border)
+    set_cell(ws1, rn, 5, '销售经理', hdr_green_font, hdr_green_fill, hdr_align, thin_border)
+    set_cell(ws1, rn, 6, '数量', hdr_green_font, hdr_green_fill, hdr_align, thin_border)
+    rn += 1
+    
+    # Collect all dept-sales-count tuples
+    items = []
+    for dept in sorted(dept_sales.keys()):
+        for sales in sorted(dept_sales[dept].keys()):
+            items.append((dept, sales, dept_sales[dept][sales]))
+    
+    # Write in two columns
+    mid = (len(items) + 1) // 2
+    max_rows = max(mid, len(items) - mid)
+    for i in range(max_rows):
+        if i < mid:
+            d1, s1, c1 = items[i]
+            set_cell(ws1, rn, 1, d1, data_font, None, data_align, thin_border)
+            set_cell(ws1, rn, 2, s1, data_font, None, data_align, thin_border)
+            set_cell(ws1, rn, 3, c1, data_font, None, data_align, thin_border)
+        else:
+            for c in [1,2,3]:
+                set_cell(ws1, rn, c, '', data_font, None, data_align, thin_border)
+        
+        if i + mid < len(items):
+            d2, s2, c2 = items[i + mid]
+            set_cell(ws1, rn, 4, d2, data_font, None, data_align, thin_border)
+            set_cell(ws1, rn, 5, s2, data_font, None, data_align, thin_border)
+            set_cell(ws1, rn, 6, c2, data_font, None, data_align, thin_border)
+        else:
+            for c in [4,5,6]:
+                set_cell(ws1, rn, c, '', data_font, None, data_align, thin_border)
+        rn += 1
+    rn += 1
+
+rn += 1
+
+# Section 3: Column source explanation
+set_cell(ws1, rn, 1, '三、里程三列数据来源说明', subtitle_font); rn += 1
+set_cell(ws1, rn, 1, '列', hdr_blue_font, hdr_blue_fill, hdr_align, thin_border)
+set_cell(ws1, rn, 2, '字段名', hdr_blue_font, hdr_blue_fill, hdr_align, thin_border)
+set_cell(ws1, rn, 3, '数据来源', hdr_blue_font, hdr_blue_fill, hdr_align, thin_border)
+set_cell(ws1, rn, 4, '说明', hdr_blue_font, hdr_blue_fill, hdr_align, thin_border)
+rn += 1
+
+col_descs = [
+    ('P', '实际行驶里程(km)', 'GPS 轨迹还原', '基于 G7 GPS 定位轨迹按考核期逐日汇总计算的运营里程（total_mileage/100000=km），本月里程考核以此列为准'),
+    ('O', '原TBOX里程(km)', '车辆 T-Box / 仪表盘读数', '当月仪表总里程（业务系统原始数据），含运营+测试+调试里程。仅干预车辆填充，非干预车辆为空'),
+    ('Q', '测试里程(km)', 'O − P（原TBOX − GPS，差额）', '仪表盘跑了但 GPS 未记录到的部分，判定为厂内测试/调试/路试等非运营里程，不计入考核。仅干预车辆填充，非干预车辆为0'),
+]
+for col_letter, name, source, desc in col_descs:
+    set_cell(ws1, rn, 1, col_letter, data_font, None, data_align, thin_border)
+    set_cell(ws1, rn, 2, name, data_font, None, Alignment(horizontal='left', vertical='center'), thin_border)
+    set_cell(ws1, rn, 3, source, data_font, None, data_align, thin_border)
+    set_cell(ws1, rn, 4, desc, data_font, None, Alignment(horizontal='left', vertical='center', wrap_text=True), thin_border)
+    ws1.row_dimensions[rn].height = 40
+    rn += 1
+
+# ===== Sheet 2: 业务考核视图 (full data,干预标黄) =====
+ws2 = wb.create_sheet('业务考核视图')
+target_order = ['车牌号','考核年份','考核月份','部门名称','销售经理','客户名称','合同编号','项目名称',
+                '考核目标ID','考核目标','考核开始日期','考核结束日期','考核天数','应考核里程(km)',
+                '原TBOX里程(km)','实际行驶里程(km)','测试里程(km)','完成率(%)','是否达标','考核状态',
+                '交车日期','预计还车日期','实际还车日期']
+
+for ci, name in enumerate(target_order, 1):
+    set_cell(ws2, 1, ci, name, hdr_blue_font, hdr_blue_fill, hdr_align, thin_border)
+
+for ri, r in enumerate(all_recs, 2):
+    is_interv = r['车牌号'] in intervention_plates
+    for ci, name in enumerate(target_order, 1):
+        val = r.get(name)
+        set_cell(ws2, ri, ci, val, data_font, yellow_fill if is_interv else None, data_align, thin_border)
+
+for col, w in enumerate([12,8,8,12,10,26,18,22,10,22,13,12,10,14,16,14,14,12,10,8,12,12,12], 1):
+    ws2.column_dimensions[get_column_letter(col)].width = w
+
+# ===== Sheet 3: 6月业务考核车辆总览 =====
+ws3 = wb.create_sheet('6月业务考核车辆总览')
+for ci, name in enumerate(target_order, 1):
+    set_cell(ws3, 1, ci, name, hdr_blue_font, hdr_blue_fill, hdr_align, thin_border)
+for ri, r in enumerate(all_recs, 2):
+    for ci, name in enumerate(target_order, 1):
+        set_cell(ws3, ri, ci, r.get(name), data_font, None, data_align, thin_border)
+for col, w in enumerate([12,8,8,12,10,26,18,22,10,22,13,12,10,14,16,14,14,12,10,8,12,12,12], 1):
+    ws3.column_dimensions[get_column_letter(col)].width = w
+
+# ===== Sheet 4: 涉及测试的业务考核车辆 =====
+ws4 = wb.create_sheet('涉及测试的业务考核车辆')
+for ci, name in enumerate(target_order, 1):
+    set_cell(ws4, 1, ci, name, hdr_blue_font, hdr_blue_fill, hdr_align, thin_border)
+for ri, r in enumerate(intervention_recs, 2):
+    for ci, name in enumerate(target_order, 1):
+        set_cell(ws4, ri, ci, r.get(name), data_font, yellow_fill, data_align, thin_border)
+for col, w in enumerate([12,8,8,12,10,26,18,22,10,22,13,12,10,14,16,14,14,12,10,8,12,12,12], 1):
+    ws4.column_dimensions[get_column_letter(col)].width = w
+
+# ========== Save ==========
+out = '租赁任务考核_2026年6月_干预情况说明.xlsx'
+wb.save(out)
+print(f'✅ {out}')
+print(f'\n概况:')
+print(f'  全量: {total_vehicles}辆, {total_records}条')
+print(f'  干预: {interv_vehicles}辆, {interv_records}条')
+print(f'  达→未达标: {flip_total}条')
+print(f'  仪表盘里程: {total_tbox:.0f} (干预{interv_tbox:.0f})')
+print(f'  GPS里程: {total_actual:.0f} (干预{interv_actual:.0f})')
+print(f'  测试里程: {total_test:.0f} (干预{interv_test:.0f})')
diff --git a/gen_resignation_jul.py b/gen_resignation_jul.py
new file mode 100644
index 0000000..66d88f2
--- /dev/null
+++ b/gen_resignation_jul.py
@@ -0,0 +1,393 @@
+#!/usr/bin/env python3
+"""生成 离职人员7月奖金处理明细 — 当月转嫁，其他不发了（18T: 1000/0/1000/1000）"""
+import os, openpyxl, warnings
+from openpyxl.styles import Font, PatternFill, Alignment, Border, Side
+from openpyxl.utils import get_column_letter
+from collections import defaultdict
+from calc_engine import *
+
+warnings.filterwarnings('ignore')
+os.chdir('/Users/kkfluous/Downloads')
+JULY_SOURCE = os.environ.get('JULY_SOURCE', '7月核算结果/租赁任务考核_2026年7月.xlsx')
+JULY_OUTPUT_DIR = os.environ.get('JULY_OUTPUT_DIR', '7月核算结果')
+
+# ── Full calc (same as main.py) ──
+print("读取源数据...")
+D = {}
+for m in range(1, 8):
+    fp = JULY_SOURCE if m == 7 else f'租赁任务考核_2026年{m}月.xlsx'
+    if os.path.exists(fp):
+        D[m] = read_file(fp, m)
+        print(f"  {m}月: {len(D[m])}条")
+
+G = {m: grp(D[m], m) for m in D}
+calc_jan_carryover(G[1])
+all_data = {}
+all_data[1] = {'当月': [{'车牌': k[0], '销售': g['销售'], '部门': g['部门'], '额': g['奖金']}
+                        for k, g in G[1].items() if g['奖金'] > 0]}
+
+for m in range(2, 8):
+    if m not in G:
+        continue
+    data_m, G[m] = calc_month(m, G, all_data, G[m-1], G[m])
+
+# ── Resignation config ──
+RESIGNATIONS = {
+    '赵连飞': {'dept': '业务二部', 'transfer_to': '刘念念'},
+    '伍仲文': {'dept': '业务六部', 'transfer_to': '钟祥'},
+    '岑彦':   {'dept': '业务六部', 'transfer_to': '钟祥'},
+    '钱婧':   {'dept': '业务三部', 'transfer_to': '—'},
+}
+resigned = set(RESIGNATIONS.keys())
+settle_month = 7
+month_data = all_data[settle_month]
+
+# ── Collect resigned payments from all_data[6] ──
+detail_records = []
+for ptype, items in month_data.items():
+    for it in items:
+        if it['销售'] not in resigned:
+            continue
+        name = it['销售']
+        amount = it['额']
+        is_dangyue = (ptype == '当月')
+        transfer_to = RESIGNATIONS[name]['transfer_to']
+        handling = f'转{transfer_to}(需确认)' if is_dangyue else '不发了'
+        transfer_disp = transfer_to if is_dangyue else '—'
+        note = '7月当月奖金，转新销售' if is_dangyue else f'{ptype}，不发了'
+
+        # Get client info from D[6]
+        client = ''
+        target_name = ''
+        for r in D.get(settle_month, []):
+            if r['车牌号'] == it['车牌'] and r['销售经理'] == name:
+                client = r.get('客户名称', '')
+                target_name = r.get('考核目标', '')
+                break
+
+        # Get group info
+        g6 = G[settle_month].get((it['车牌'], name))
+        monthly_bonus = g6['奖励达标'] if g6 else 0
+        days = g6['天数'] if g6 else ''
+
+        # 盈亏查询月
+        if ptype == '当月' or ptype == f'累计补发{settle_month}月':
+            loss_month = settle_month
+        elif ptype == '结转':
+            loss_month = settle_month
+        elif ptype.startswith('补发'):
+            try:
+                loss_month = int(ptype.replace('补发', '').replace('月', ''))
+            except:
+                loss_month = settle_month
+        else:
+            loss_month = settle_month
+
+        detail_records.append({
+            'name': name, 'plate': it['车牌'], 'pay_type': ptype,
+            'amount': amount, 'handling': handling, 'transfer': transfer_disp,
+            'note': note, 'client': client, 'loss_month': loss_month,
+            'loss_status': '否', 'monthly_bonus': monthly_bonus,
+            'days': days, 'target': target_name,
+        })
+
+print(f"离职人员发放记录: {len(detail_records)} 条")
+
+# ── Styles ──
+HEAD_FILL = PatternFill('solid', fgColor='4472C4')
+HEAD_FONT = Font(bold=True, color='FFFFFF', size=10)
+SUB_FILL  = PatternFill('solid', fgColor='D9E1F2')
+ALT_FILL  = PatternFill('solid', fgColor='F2F2F2')
+TITLE_FNT = Font(bold=True, size=14)
+NOTE_FNT  = Font(size=10, color='C00000')
+NORM_FNT  = Font(size=10)
+BOLD_FNT  = Font(bold=True, size=10)
+THIN = Side('thin', color='B4B4B4')
+BORD = Border(THIN, THIN, THIN, THIN)
+CC = Alignment(horizontal='center', vertical='center', wrap_text=True)
+LL = Alignment(horizontal='left', vertical='center', wrap_text=True)
+
+def hdr(ws, r, nc):
+    for c in range(1, nc+1):
+        cell = ws.cell(row=r, column=c)
+        cell.fill = HEAD_FILL; cell.font = HEAD_FONT; cell.alignment = CC; cell.border = BORD
+
+def drw(ws, r, nc, alt=False):
+    for c in range(1, nc+1):
+        cell = ws.cell(row=r, column=c)
+        cell.font = NORM_FNT; cell.alignment = CC; cell.border = BORD
+        if alt: cell.fill = ALT_FILL
+
+wb = openpyxl.Workbook()
+
+# ═══════════════════════════════════════
+# Sheet 1: 离职人员发放处理总览
+# ═══════════════════════════════════════
+ws1 = wb.active
+ws1.title = '离职人员发放处理总览'
+
+ws1.merge_cells('A1:G1')
+ws1.cell(row=1, column=1, value='2026年7月核算 — 离职人员奖金处理方案').font = TITLE_FNT
+ws1.cell(row=1, column=1).alignment = CC
+ws1.row_dimensions[1].height = 30
+
+ws1.merge_cells('A2:G2')
+ws1.cell(row=2, column=1,
+    value='处理规则：仅7月当月奖金转给新接手的销售（需与其他部门确认是否发放）；其余结转、补发、累计补发等一律不发了。').font = NOTE_FNT
+ws1.cell(row=2, column=1).alignment = LL
+ws1.row_dimensions[2].height = 22
+
+r = 4
+oh = ['离职人员', '部门', '业务转嫁至', '不发了', '转新销售(需确认)', '原考核应发合计']
+for c, h in enumerate(oh, 1):
+    ws1.cell(row=r, column=c, value=h)
+hdr(ws1, r, 6)
+r += 1
+
+total_not = 0; total_trf = 0
+for name in ['赵连飞', '伍仲文', '岑彦', '钱婧']:
+    info = RESIGNATIONS[name]
+    recs = [p for p in detail_records if p['name'] == name]
+    not_pay = sum(p['amount'] for p in recs if p['pay_type'] != '当月')
+    transfer = sum(p['amount'] for p in recs if p['pay_type'] == '当月')
+    total_not += not_pay; total_trf += transfer
+    ws1.cell(row=r, column=1, value=name)
+    ws1.cell(row=r, column=2, value=info['dept'])
+    ws1.cell(row=r, column=3, value=info['transfer_to'])
+    ws1.cell(row=r, column=4, value=round(not_pay, 2))
+    ws1.cell(row=r, column=5, value=round(transfer, 2))
+    ws1.cell(row=r, column=6, value=round(not_pay+transfer, 2))
+    drw(ws1, r, 6)
+    ws1.row_dimensions[r].height = 22
+    r += 1
+
+# 合计
+for c in range(1, 7):
+    ws1.cell(row=r, column=c).fill = SUB_FILL
+    ws1.cell(row=r, column=c).border = BORD
+    ws1.cell(row=r, column=c).font = BOLD_FNT
+    ws1.cell(row=r, column=c).alignment = CC
+ws1.cell(row=r, column=1, value='合计')
+ws1.cell(row=r, column=4, value=round(total_not, 2))
+ws1.cell(row=r, column=5, value=round(total_trf, 2))
+ws1.cell(row=r, column=6, value=round(total_not+total_trf, 2))
+
+r += 2
+ws1.merge_cells(start_row=r, start_column=1, end_row=r, end_column=6)
+ws1.cell(row=r, column=1, value='转嫁接收方汇总（需与其他部门确认是否发放）').font = Font(bold=True, size=12, color='1F4E78')
+ws1.cell(row=r, column=1).alignment = CC
+r += 1
+
+for c, hh in enumerate(['接收人', '来源离职人员', '转嫁金额', '备注'], 1):
+    ws1.cell(row=r, column=c, value=hh)
+hdr(ws1, r, 4)
+r += 1
+
+by_recv = defaultdict(lambda: {'sources':[], 'amt':0})
+for rec in detail_records:
+    if rec['pay_type'] == '当月':
+        t = rec['transfer']
+        by_recv[t]['sources'].append(rec['name'])
+        by_recv[t]['amt'] += rec['amount']
+
+for receiver in ['刘念念', '钟祥']:
+    d = by_recv[receiver]
+    ws1.cell(row=r, column=1, value=receiver)
+    ws1.cell(row=r, column=2, value='、'.join(sorted(set(d['sources']))))
+    ws1.cell(row=r, column=3, value=round(d['amt'], 2))
+    ws1.cell(row=r, column=4, value='需与相关部门确认是否发放给新销售')
+    drw(ws1, r, 4)
+    r += 1
+
+for i, w in enumerate([14, 22, 18, 18, 18, 18], 1):
+    ws1.column_dimensions[get_column_letter(i)].width = w
+
+# ═══════════════════════════════════════
+# Sheet 2: 发放明细(逐条)
+# ═══════════════════════════════════════
+ws2 = wb.create_sheet('发放明细(逐条)')
+ws2.merge_cells('A1:M1')
+ws2.cell(row=1, column=1, value='离职人员 — 7月考核应发逐条明细').font = TITLE_FNT
+ws2.cell(row=1, column=1).alignment = CC
+ws2.row_dimensions[1].height = 28
+
+dh = ['离职人员', '车牌号', '发放类型', '考核应发', '盈亏查询月', '客户名称',
+      '客户盈亏', '处理方式', '转嫁接收人', '7月考核天数', '月度奖励', '考核目标', '备注']
+r2 = 3
+for c, h in enumerate(dh, 1):
+    ws2.cell(row=r2, column=c, value=h)
+hdr(ws2, r2, 13)
+r2 += 1
+
+name_order = ['赵连飞', '伍仲文', '岑彦', '钱婧']
+detail_records.sort(key=lambda x: (name_order.index(x['name']), 0 if x['pay_type']=='当月' else 1, x['plate']))
+
+for i, rec in enumerate(detail_records):
+    ws2.cell(row=r2, column=1, value=rec['name'])
+    ws2.cell(row=r2, column=2, value=rec['plate'])
+    ws2.cell(row=r2, column=3, value=rec['pay_type'])
+    ws2.cell(row=r2, column=4, value=round(rec['amount'], 2))
+    ws2.cell(row=r2, column=5, value=f"{rec['loss_month']}月")
+    ws2.cell(row=r2, column=6, value=rec['client'])
+    ws2.cell(row=r2, column=7, value=rec['loss_status'])
+    ws2.cell(row=r2, column=8, value=rec['handling'])
+    ws2.cell(row=r2, column=9, value=rec['transfer'])
+    ws2.cell(row=r2, column=10, value=rec['days'] if rec['days'] else '')
+    ws2.cell(row=r2, column=11, value=rec['monthly_bonus'] if rec['monthly_bonus'] else '')
+    ws2.cell(row=r2, column=12, value=rec['target'])
+    ws2.cell(row=r2, column=13, value=rec['note'])
+    drw(ws2, r2, 13, alt=(i%2==1))
+    ws2.row_dimensions[r2].height = 22
+    r2 += 1
+
+# 合计行
+ws2.cell(row=r2, column=3, value='合计')
+ws2.cell(row=r2, column=4, value=round(sum(r['amount'] for r in detail_records), 2))
+for c in range(1, 14):
+    ws2.cell(row=r2, column=c).fill = SUB_FILL
+    ws2.cell(row=r2, column=c).border = BORD
+    ws2.cell(row=r2, column=c).font = BOLD_FNT
+    ws2.cell(row=r2, column=c).alignment = CC
+
+for i, w in enumerate([10, 12, 14, 14, 10, 24, 8, 18, 14, 12, 10, 22, 28], 1):
+    ws2.column_dimensions[get_column_letter(i)].width = w
+
+# ═══════════════════════════════════════
+# Sheets 3-5: 个人明细
+# ═══════════════════════════════════════
+for name, title_txt, transfer_to in [
+    ('赵连飞', '赵连飞（2026-04-30离职，业务转→刘念念）', '刘念念'),
+    ('伍仲文', '伍仲文（2026-05-09离职，业务转→钟祥）', '钟祥'),
+    ('岑彦', '岑彦（2026-05-09离职，业务转→钟祥）', '钟祥'),
+    ('钱婧', '钱婧（已离职，所有奖金不再发放）', '—'),
+]:
+    ws = wb.create_sheet(name)
+    ws.merge_cells('A1:K1')
+    ws.cell(row=1, column=1, value=title_txt).font = TITLE_FNT
+    ws.cell(row=1, column=1).alignment = CC
+    ws.row_dimensions[1].height = 28
+
+    ph = ['车牌号', '发放类型', '考核应发', '盈亏查询月', '客户名称', '客户盈亏',
+          '处理方式', '7月考核天数', '月度奖励', '考核目标', '备注']
+    r3 = 3
+    for c, h in enumerate(ph, 1):
+        ws.cell(row=r3, column=c, value=h)
+    hdr(ws, r3, 11)
+    r3 += 1
+
+    person_recs = [p for p in detail_records if p['name'] == name]
+    dangyue_recs = [p for p in person_recs if p['pay_type'] == '当月']
+    other_recs = [p for p in person_recs if p['pay_type'] != '当月']
+
+    for i, rec in enumerate(person_recs):
+        h_disp = rec['handling']
+        ws.cell(row=r3, column=1, value=rec['plate'])
+        ws.cell(row=r3, column=2, value=rec['pay_type'])
+        ws.cell(row=r3, column=3, value=round(rec['amount'], 2))
+        ws.cell(row=r3, column=4, value=f"{rec['loss_month']}月")
+        ws.cell(row=r3, column=5, value=rec['client'])
+        ws.cell(row=r3, column=6, value=rec['loss_status'])
+        ws.cell(row=r3, column=7, value=h_disp)
+        ws.cell(row=r3, column=8, value=rec['days'] if rec['days'] else '')
+        ws.cell(row=r3, column=9, value=rec['monthly_bonus'] if rec['monthly_bonus'] else '')
+        ws.cell(row=r3, column=10, value=rec['target'])
+        ws.cell(row=r3, column=11, value=rec['note'])
+        drw(ws, r3, 11, alt=(i%2==1))
+        ws.row_dimensions[r3].height = 22
+        r3 += 1
+
+    dn = sum(p['amount'] for p in dangyue_recs)
+    ot = sum(p['amount'] for p in other_recs)
+    gt = dn + ot
+
+    # 不发了小计
+    ws.cell(row=r3, column=2, value='不发了小计').font = BOLD_FNT
+    ws.cell(row=r3, column=3, value=round(ot, 2)).font = BOLD_FNT
+    for c in range(1, 12):
+        ws.cell(row=r3, column=c).fill = SUB_FILL; ws.cell(row=r3, column=c).border = BORD
+    ws.cell(row=r3, column=2).alignment = CC; ws.cell(row=r3, column=3).alignment = CC
+    r3 += 1
+
+    # 转嫁小计
+    ws.cell(row=r3, column=2, value=f'转{transfer_to}小计').font = BOLD_FNT
+    ws.cell(row=r3, column=3, value=round(dn, 2)).font = BOLD_FNT
+    ws.cell(row=r3, column=11, value='需与其他部门确认').font = Font(size=9, color='C00000')
+    for c in range(1, 12):
+        ws.cell(row=r3, column=c).fill = SUB_FILL; ws.cell(row=r3, column=c).border = BORD
+    ws.cell(row=r3, column=2).alignment = CC; ws.cell(row=r3, column=3).alignment = CC
+    r3 += 1
+
+    # 合计
+    ws.cell(row=r3, column=2, value='合计').font = BOLD_FNT
+    ws.cell(row=r3, column=3, value=round(gt, 2)).font = BOLD_FNT
+    for c in range(1, 12):
+        ws.cell(row=r3, column=c).fill = SUB_FILL; ws.cell(row=r3, column=c).border = BORD
+    ws.cell(row=r3, column=2).alignment = CC; ws.cell(row=r3, column=3).alignment = CC
+
+    for i, w in enumerate([12, 14, 14, 10, 26, 8, 18, 12, 10, 22, 30], 1):
+        ws.column_dimensions[get_column_letter(i)].width = w
+
+# ═══════════════════════════════════════
+# Sheets 6-7: 转嫁明细
+# ═══════════════════════════════════════
+for title_txt, sources, receiver in [
+    ('转嫁至刘念念的奖金 — 来源：赵连飞(2026-04-30离职)', ['赵连飞'], '刘念念'),
+    ('转嫁至钟祥的奖金 — 来源：伍仲文(2026-05-09离职)、岑彦(2026-05-09离职)', ['伍仲文', '岑彦'], '钟祥'),
+]:
+    ws = wb.create_sheet(f'转{receiver}明细')
+    ws.merge_cells('A1:K1')
+    ws.cell(row=1, column=1, value=title_txt).font = TITLE_FNT
+    ws.cell(row=1, column=1).alignment = CC
+    ws.row_dimensions[1].height = 28
+
+    ws.merge_cells('A2:K2')
+    ws.cell(row=2, column=1, value='⚠ 以下金额需与其他部门确认是否可发放给新销售').font = NOTE_FNT
+    ws.cell(row=2, column=1).alignment = LL
+
+    th = ['来源离职人员', '车牌号', '发放类型', '考核应发', '盈亏查询月', '客户名称',
+          '客户盈亏', '7月考核天数', '月度奖励', '考核目标', '备注']
+    r4 = 4
+    for c, h in enumerate(th, 1):
+        ws.cell(row=r4, column=c, value=h)
+    hdr(ws, r4, 11)
+    r4 += 1
+
+    trf_recs = [p for p in detail_records if p['name'] in sources and p['pay_type'] == '当月']
+    for i, rec in enumerate(trf_recs):
+        ws.cell(row=r4, column=1, value=rec['name'])
+        ws.cell(row=r4, column=2, value=rec['plate'])
+        ws.cell(row=r4, column=3, value='当月')
+        ws.cell(row=r4, column=4, value=round(rec['amount'], 2))
+        ws.cell(row=r4, column=5, value=f"{rec['loss_month']}月")
+        ws.cell(row=r4, column=6, value=rec['client'])
+        ws.cell(row=r4, column=7, value=rec['loss_status'])
+        ws.cell(row=r4, column=8, value=rec['days'] if rec['days'] else '')
+        ws.cell(row=r4, column=9, value=rec['monthly_bonus'] if rec['monthly_bonus'] else '')
+        ws.cell(row=r4, column=10, value=rec['target'])
+        ws.cell(row=r4, column=11, value='需确认是否发给新销售')
+        drw(ws, r4, 11, alt=(i%2==1))
+        ws.row_dimensions[r4].height = 22
+        r4 += 1
+
+    ws.cell(row=r4, column=3, value='合计')
+    ws.cell(row=r4, column=4, value=round(sum(r['amount'] for r in trf_recs), 2))
+    for c in range(1, 12):
+        ws.cell(row=r4, column=c).fill = SUB_FILL; ws.cell(row=r4, column=c).border = BORD
+        ws.cell(row=r4, column=c).font = BOLD_FNT; ws.cell(row=r4, column=c).alignment = CC
+
+    for i, w in enumerate([14, 12, 14, 14, 10, 26, 8, 12, 10, 22, 22], 1):
+        ws.column_dimensions[get_column_letter(i)].width = w
+
+# ── Save ──
+os.makedirs(JULY_OUTPUT_DIR, exist_ok=True)
+out = os.path.join(JULY_OUTPUT_DIR, '离职人员7月奖金处理明细.xlsx')
+wb.save(out)
+
+print(f'\n✅ {out} ({len(wb.sheetnames)} sheets)')
+print(f'\n汇总:')
+for name in ['赵连飞', '伍仲文', '岑彦', '钱婧']:
+    recs = [p for p in detail_records if p['name'] == name]
+    not_pay = sum(p['amount'] for p in recs if p['pay_type'] != '当月')
+    trf = sum(p['amount'] for p in recs if p['pay_type'] == '当月')
+    print(f'  {name}: 不发了={not_pay:.2f}  转{RESIGNATIONS[name]["transfer_to"]}={trf:.2f}  合计={not_pay+trf:.2f}')
+print(f'  总计: 不发了={total_not:.2f}  转新销售={total_trf:.2f}')
diff --git a/gen_resignation_jun.py b/gen_resignation_jun.py
new file mode 100644
index 0000000..af90d67
--- /dev/null
+++ b/gen_resignation_jun.py
@@ -0,0 +1,388 @@
+#!/usr/bin/env python3
+"""生成 离职人员6月奖金处理明细 — 当月转嫁，其他不发了（18T: 1000/0/1000/1000）"""
+import os, openpyxl, warnings
+from openpyxl.styles import Font, PatternFill, Alignment, Border, Side
+from openpyxl.utils import get_column_letter
+from collections import defaultdict
+from calc_engine import *
+
+warnings.filterwarnings('ignore')
+os.chdir('/Users/kkfluous/Downloads')
+
+# ── Full calc (same as main.py) ──
+print("读取源数据...")
+D = {}
+for m in range(1, 7):
+    fp = f'租赁任务考核_2026年{m}月.xlsx'
+    if os.path.exists(fp):
+        D[m] = read_file(fp, m)
+        print(f"  {m}月: {len(D[m])}条")
+
+G = {m: grp(D[m], m) for m in D}
+calc_jan_carryover(G[1])
+all_data = {}
+all_data[1] = {'当月': [{'车牌': k[0], '销售': g['销售'], '部门': g['部门'], '额': g['奖金']}
+                        for k, g in G[1].items() if g['奖金'] > 0]}
+
+for m in range(2, 7):
+    if m not in G:
+        continue
+    data_m, G[m] = calc_month(m, G, all_data, G[m-1], G[m])
+
+# ── Resignation config ──
+RESIGNATIONS = {
+    '赵连飞': {'dept': '业务二部', 'transfer_to': '刘念念'},
+    '伍仲文': {'dept': '业务六部', 'transfer_to': '钟祥'},
+    '岑彦':   {'dept': '业务六部', 'transfer_to': '钟祥'},
+}
+resigned = set(RESIGNATIONS.keys())
+settle_month = 6
+month_data = all_data[settle_month]
+
+# ── Collect resigned payments from all_data[6] ──
+detail_records = []
+for ptype, items in month_data.items():
+    for it in items:
+        if it['销售'] not in resigned:
+            continue
+        name = it['销售']
+        amount = it['额']
+        is_dangyue = (ptype == '当月')
+        transfer_to = RESIGNATIONS[name]['transfer_to']
+        handling = f'转{transfer_to}(需确认)' if is_dangyue else '不发了'
+        transfer_disp = transfer_to if is_dangyue else '—'
+        note = '6月当月达标，转新销售' if is_dangyue else f'{ptype}，不发了'
+
+        # Get client info from D[6]
+        client = ''
+        target_name = ''
+        for r in D.get(settle_month, []):
+            if r['车牌号'] == it['车牌'] and r['销售经理'] == name:
+                client = r.get('客户名称', '')
+                target_name = r.get('考核目标', '')
+                break
+
+        # Get group info
+        g6 = G[settle_month].get((it['车牌'], name))
+        monthly_bonus = g6['奖励达标'] if g6 else 0
+        days = g6['天数'] if g6 else ''
+
+        # 盈亏查询月
+        if ptype == '当月' or ptype == f'累计补发{settle_month}月':
+            loss_month = settle_month
+        elif ptype == '结转':
+            loss_month = settle_month
+        elif ptype.startswith('补发'):
+            try:
+                loss_month = int(ptype.replace('补发', '').replace('月', ''))
+            except:
+                loss_month = settle_month
+        else:
+            loss_month = settle_month
+
+        detail_records.append({
+            'name': name, 'plate': it['车牌'], 'pay_type': ptype,
+            'amount': amount, 'handling': handling, 'transfer': transfer_disp,
+            'note': note, 'client': client, 'loss_month': loss_month,
+            'loss_status': '否', 'monthly_bonus': monthly_bonus,
+            'days': days, 'target': target_name,
+        })
+
+print(f"离职人员发放记录: {len(detail_records)} 条")
+
+# ── Styles ──
+HEAD_FILL = PatternFill('solid', fgColor='4472C4')
+HEAD_FONT = Font(bold=True, color='FFFFFF', size=10)
+SUB_FILL  = PatternFill('solid', fgColor='D9E1F2')
+ALT_FILL  = PatternFill('solid', fgColor='F2F2F2')
+TITLE_FNT = Font(bold=True, size=14)
+NOTE_FNT  = Font(size=10, color='C00000')
+NORM_FNT  = Font(size=10)
+BOLD_FNT  = Font(bold=True, size=10)
+THIN = Side('thin', color='B4B4B4')
+BORD = Border(THIN, THIN, THIN, THIN)
+CC = Alignment(horizontal='center', vertical='center', wrap_text=True)
+LL = Alignment(horizontal='left', vertical='center', wrap_text=True)
+
+def hdr(ws, r, nc):
+    for c in range(1, nc+1):
+        cell = ws.cell(row=r, column=c)
+        cell.fill = HEAD_FILL; cell.font = HEAD_FONT; cell.alignment = CC; cell.border = BORD
+
+def drw(ws, r, nc, alt=False):
+    for c in range(1, nc+1):
+        cell = ws.cell(row=r, column=c)
+        cell.font = NORM_FNT; cell.alignment = CC; cell.border = BORD
+        if alt: cell.fill = ALT_FILL
+
+wb = openpyxl.Workbook()
+
+# ═══════════════════════════════════════
+# Sheet 1: 离职人员发放处理总览
+# ═══════════════════════════════════════
+ws1 = wb.active
+ws1.title = '离职人员发放处理总览'
+
+ws1.merge_cells('A1:G1')
+ws1.cell(row=1, column=1, value='2026年6月核算 — 离职人员奖金处理方案').font = TITLE_FNT
+ws1.cell(row=1, column=1).alignment = CC
+ws1.row_dimensions[1].height = 30
+
+ws1.merge_cells('A2:G2')
+ws1.cell(row=2, column=1,
+    value='处理规则：仅6月当月达标奖金转给新接手的销售（需与其他部门确认是否发放）；其余结转、补发、累计补发等一律不发了。').font = NOTE_FNT
+ws1.cell(row=2, column=1).alignment = LL
+ws1.row_dimensions[2].height = 22
+
+r = 4
+oh = ['离职人员', '部门', '业务转嫁至', '不发了', '转新销售(需确认)', '原考核应发合计']
+for c, h in enumerate(oh, 1):
+    ws1.cell(row=r, column=c, value=h)
+hdr(ws1, r, 6)
+r += 1
+
+total_not = 0; total_trf = 0
+for name in ['赵连飞', '伍仲文', '岑彦']:
+    info = RESIGNATIONS[name]
+    recs = [p for p in detail_records if p['name'] == name]
+    not_pay = sum(p['amount'] for p in recs if p['pay_type'] != '当月')
+    transfer = sum(p['amount'] for p in recs if p['pay_type'] == '当月')
+    total_not += not_pay; total_trf += transfer
+    ws1.cell(row=r, column=1, value=name)
+    ws1.cell(row=r, column=2, value=info['dept'])
+    ws1.cell(row=r, column=3, value=info['transfer_to'])
+    ws1.cell(row=r, column=4, value=round(not_pay, 2))
+    ws1.cell(row=r, column=5, value=round(transfer, 2))
+    ws1.cell(row=r, column=6, value=round(not_pay+transfer, 2))
+    drw(ws1, r, 6)
+    ws1.row_dimensions[r].height = 22
+    r += 1
+
+# 合计
+for c in range(1, 7):
+    ws1.cell(row=r, column=c).fill = SUB_FILL
+    ws1.cell(row=r, column=c).border = BORD
+    ws1.cell(row=r, column=c).font = BOLD_FNT
+    ws1.cell(row=r, column=c).alignment = CC
+ws1.cell(row=r, column=1, value='合计')
+ws1.cell(row=r, column=4, value=round(total_not, 2))
+ws1.cell(row=r, column=5, value=round(total_trf, 2))
+ws1.cell(row=r, column=6, value=round(total_not+total_trf, 2))
+
+r += 2
+ws1.merge_cells(start_row=r, start_column=1, end_row=r, end_column=6)
+ws1.cell(row=r, column=1, value='转嫁接收方汇总（需与其他部门确认是否发放）').font = Font(bold=True, size=12, color='1F4E78')
+ws1.cell(row=r, column=1).alignment = CC
+r += 1
+
+for c, hh in enumerate(['接收人', '来源离职人员', '转嫁金额', '备注'], 1):
+    ws1.cell(row=r, column=c, value=hh)
+hdr(ws1, r, 4)
+r += 1
+
+by_recv = defaultdict(lambda: {'sources':[], 'amt':0})
+for rec in detail_records:
+    if rec['pay_type'] == '当月':
+        t = rec['transfer']
+        by_recv[t]['sources'].append(rec['name'])
+        by_recv[t]['amt'] += rec['amount']
+
+for receiver in ['刘念念', '钟祥']:
+    d = by_recv[receiver]
+    ws1.cell(row=r, column=1, value=receiver)
+    ws1.cell(row=r, column=2, value='、'.join(sorted(set(d['sources']))))
+    ws1.cell(row=r, column=3, value=round(d['amt'], 2))
+    ws1.cell(row=r, column=4, value='需与相关部门确认是否发放给新销售')
+    drw(ws1, r, 4)
+    r += 1
+
+for i, w in enumerate([14, 22, 18, 18, 18, 18], 1):
+    ws1.column_dimensions[get_column_letter(i)].width = w
+
+# ═══════════════════════════════════════
+# Sheet 2: 发放明细(逐条)
+# ═══════════════════════════════════════
+ws2 = wb.create_sheet('发放明细(逐条)')
+ws2.merge_cells('A1:M1')
+ws2.cell(row=1, column=1, value='离职人员 — 6月考核应发逐条明细').font = TITLE_FNT
+ws2.cell(row=1, column=1).alignment = CC
+ws2.row_dimensions[1].height = 28
+
+dh = ['离职人员', '车牌号', '发放类型', '考核应发', '盈亏查询月', '客户名称',
+      '客户盈亏', '处理方式', '转嫁接收人', '6月考核天数', '月度奖励', '考核目标', '备注']
+r2 = 3
+for c, h in enumerate(dh, 1):
+    ws2.cell(row=r2, column=c, value=h)
+hdr(ws2, r2, 13)
+r2 += 1
+
+name_order = ['赵连飞', '伍仲文', '岑彦']
+detail_records.sort(key=lambda x: (name_order.index(x['name']), 0 if x['pay_type']=='当月' else 1, x['plate']))
+
+for i, rec in enumerate(detail_records):
+    ws2.cell(row=r2, column=1, value=rec['name'])
+    ws2.cell(row=r2, column=2, value=rec['plate'])
+    ws2.cell(row=r2, column=3, value=rec['pay_type'])
+    ws2.cell(row=r2, column=4, value=round(rec['amount'], 2))
+    ws2.cell(row=r2, column=5, value=f"{rec['loss_month']}月")
+    ws2.cell(row=r2, column=6, value=rec['client'])
+    ws2.cell(row=r2, column=7, value=rec['loss_status'])
+    ws2.cell(row=r2, column=8, value=rec['handling'])
+    ws2.cell(row=r2, column=9, value=rec['transfer'])
+    ws2.cell(row=r2, column=10, value=rec['days'] if rec['days'] else '')
+    ws2.cell(row=r2, column=11, value=rec['monthly_bonus'] if rec['monthly_bonus'] else '')
+    ws2.cell(row=r2, column=12, value=rec['target'])
+    ws2.cell(row=r2, column=13, value=rec['note'])
+    drw(ws2, r2, 13, alt=(i%2==1))
+    ws2.row_dimensions[r2].height = 22
+    r2 += 1
+
+# 合计行
+ws2.cell(row=r2, column=3, value='合计')
+ws2.cell(row=r2, column=4, value=round(sum(r['amount'] for r in detail_records), 2))
+for c in range(1, 14):
+    ws2.cell(row=r2, column=c).fill = SUB_FILL
+    ws2.cell(row=r2, column=c).border = BORD
+    ws2.cell(row=r2, column=c).font = BOLD_FNT
+    ws2.cell(row=r2, column=c).alignment = CC
+
+for i, w in enumerate([10, 12, 14, 14, 10, 24, 8, 18, 14, 12, 10, 22, 28], 1):
+    ws2.column_dimensions[get_column_letter(i)].width = w
+
+# ═══════════════════════════════════════
+# Sheets 3-5: 个人明细
+# ═══════════════════════════════════════
+for name, title_txt, transfer_to in [
+    ('赵连飞', '赵连飞（2026-04-30离职，业务转→刘念念）', '刘念念'),
+    ('伍仲文', '伍仲文（2026-05-09离职，业务转→钟祥）', '钟祥'),
+    ('岑彦', '岑彦（2026-05-09离职，业务转→钟祥）', '钟祥'),
+]:
+    ws = wb.create_sheet(name)
+    ws.merge_cells('A1:K1')
+    ws.cell(row=1, column=1, value=title_txt).font = TITLE_FNT
+    ws.cell(row=1, column=1).alignment = CC
+    ws.row_dimensions[1].height = 28
+
+    ph = ['车牌号', '发放类型', '考核应发', '盈亏查询月', '客户名称', '客户盈亏',
+          '处理方式', '6月考核天数', '月度奖励', '考核目标', '备注']
+    r3 = 3
+    for c, h in enumerate(ph, 1):
+        ws.cell(row=r3, column=c, value=h)
+    hdr(ws, r3, 11)
+    r3 += 1
+
+    person_recs = [p for p in detail_records if p['name'] == name]
+    dangyue_recs = [p for p in person_recs if p['pay_type'] == '当月']
+    other_recs = [p for p in person_recs if p['pay_type'] != '当月']
+
+    for i, rec in enumerate(person_recs):
+        h_disp = rec['handling']
+        ws.cell(row=r3, column=1, value=rec['plate'])
+        ws.cell(row=r3, column=2, value=rec['pay_type'])
+        ws.cell(row=r3, column=3, value=round(rec['amount'], 2))
+        ws.cell(row=r3, column=4, value=f"{rec['loss_month']}月")
+        ws.cell(row=r3, column=5, value=rec['client'])
+        ws.cell(row=r3, column=6, value=rec['loss_status'])
+        ws.cell(row=r3, column=7, value=h_disp)
+        ws.cell(row=r3, column=8, value=rec['days'] if rec['days'] else '')
+        ws.cell(row=r3, column=9, value=rec['monthly_bonus'] if rec['monthly_bonus'] else '')
+        ws.cell(row=r3, column=10, value=rec['target'])
+        ws.cell(row=r3, column=11, value=rec['note'])
+        drw(ws, r3, 11, alt=(i%2==1))
+        ws.row_dimensions[r3].height = 22
+        r3 += 1
+
+    dn = sum(p['amount'] for p in dangyue_recs)
+    ot = sum(p['amount'] for p in other_recs)
+    gt = dn + ot
+
+    # 不发了小计
+    ws.cell(row=r3, column=2, value='不发了小计').font = BOLD_FNT
+    ws.cell(row=r3, column=3, value=round(ot, 2)).font = BOLD_FNT
+    for c in range(1, 12):
+        ws.cell(row=r3, column=c).fill = SUB_FILL; ws.cell(row=r3, column=c).border = BORD
+    ws.cell(row=r3, column=2).alignment = CC; ws.cell(row=r3, column=3).alignment = CC
+    r3 += 1
+
+    # 转嫁小计
+    ws.cell(row=r3, column=2, value=f'转{transfer_to}小计').font = BOLD_FNT
+    ws.cell(row=r3, column=3, value=round(dn, 2)).font = BOLD_FNT
+    ws.cell(row=r3, column=11, value='需与其他部门确认').font = Font(size=9, color='C00000')
+    for c in range(1, 12):
+        ws.cell(row=r3, column=c).fill = SUB_FILL; ws.cell(row=r3, column=c).border = BORD
+    ws.cell(row=r3, column=2).alignment = CC; ws.cell(row=r3, column=3).alignment = CC
+    r3 += 1
+
+    # 合计
+    ws.cell(row=r3, column=2, value='合计').font = BOLD_FNT
+    ws.cell(row=r3, column=3, value=round(gt, 2)).font = BOLD_FNT
+    for c in range(1, 12):
+        ws.cell(row=r3, column=c).fill = SUB_FILL; ws.cell(row=r3, column=c).border = BORD
+    ws.cell(row=r3, column=2).alignment = CC; ws.cell(row=r3, column=3).alignment = CC
+
+    for i, w in enumerate([12, 14, 14, 10, 26, 8, 18, 12, 10, 22, 30], 1):
+        ws.column_dimensions[get_column_letter(i)].width = w
+
+# ═══════════════════════════════════════
+# Sheets 6-7: 转嫁明细
+# ═══════════════════════════════════════
+for title_txt, sources, receiver in [
+    ('转嫁至刘念念的奖金 — 来源：赵连飞(2026-04-30离职)', ['赵连飞'], '刘念念'),
+    ('转嫁至钟祥的奖金 — 来源：伍仲文(2026-05-09离职)、岑彦(2026-05-09离职)', ['伍仲文', '岑彦'], '钟祥'),
+]:
+    ws = wb.create_sheet(f'转{receiver}明细')
+    ws.merge_cells('A1:K1')
+    ws.cell(row=1, column=1, value=title_txt).font = TITLE_FNT
+    ws.cell(row=1, column=1).alignment = CC
+    ws.row_dimensions[1].height = 28
+
+    ws.merge_cells('A2:K2')
+    ws.cell(row=2, column=1, value='⚠ 以下金额需与其他部门确认是否可发放给新销售').font = NOTE_FNT
+    ws.cell(row=2, column=1).alignment = LL
+
+    th = ['来源离职人员', '车牌号', '发放类型', '考核应发', '盈亏查询月', '客户名称',
+          '客户盈亏', '6月考核天数', '月度奖励', '考核目标', '备注']
+    r4 = 4
+    for c, h in enumerate(th, 1):
+        ws.cell(row=r4, column=c, value=h)
+    hdr(ws, r4, 11)
+    r4 += 1
+
+    trf_recs = [p for p in detail_records if p['name'] in sources and p['pay_type'] == '当月']
+    for i, rec in enumerate(trf_recs):
+        ws.cell(row=r4, column=1, value=rec['name'])
+        ws.cell(row=r4, column=2, value=rec['plate'])
+        ws.cell(row=r4, column=3, value='当月')
+        ws.cell(row=r4, column=4, value=round(rec['amount'], 2))
+        ws.cell(row=r4, column=5, value=f"{rec['loss_month']}月")
+        ws.cell(row=r4, column=6, value=rec['client'])
+        ws.cell(row=r4, column=7, value=rec['loss_status'])
+        ws.cell(row=r4, column=8, value=rec['days'] if rec['days'] else '')
+        ws.cell(row=r4, column=9, value=rec['monthly_bonus'] if rec['monthly_bonus'] else '')
+        ws.cell(row=r4, column=10, value=rec['target'])
+        ws.cell(row=r4, column=11, value='需确认是否发给新销售')
+        drw(ws, r4, 11, alt=(i%2==1))
+        ws.row_dimensions[r4].height = 22
+        r4 += 1
+
+    ws.cell(row=r4, column=3, value='合计')
+    ws.cell(row=r4, column=4, value=round(sum(r['amount'] for r in trf_recs), 2))
+    for c in range(1, 12):
+        ws.cell(row=r4, column=c).fill = SUB_FILL; ws.cell(row=r4, column=c).border = BORD
+        ws.cell(row=r4, column=c).font = BOLD_FNT; ws.cell(row=r4, column=c).alignment = CC
+
+    for i, w in enumerate([14, 12, 14, 14, 10, 26, 8, 12, 10, 22, 22], 1):
+        ws.column_dimensions[get_column_letter(i)].width = w
+
+# ── Save ──
+out = '离职人员6月奖金处理明细.xlsx'
+wb.save(out)
+
+print(f'\n✅ {out} ({len(wb.sheetnames)} sheets)')
+print(f'\n汇总:')
+for name in ['赵连飞', '伍仲文', '岑彦']:
+    recs = [p for p in detail_records if p['name'] == name]
+    not_pay = sum(p['amount'] for p in recs if p['pay_type'] != '当月')
+    trf = sum(p['amount'] for p in recs if p['pay_type'] == '当月')
+    print(f'  {name}: 不发了={not_pay:.2f}  转{RESIGNATIONS[name]["transfer_to"]}={trf:.2f}  合计={not_pay+trf:.2f}')
+print(f'  总计: 不发了={total_not:.2f}  转新销售={total_trf:.2f}')
diff --git a/gps_replace_full.py b/gps_replace_full.py
new file mode 100644
index 0000000..3d99b52
--- /dev/null
+++ b/gps_replace_full.py
@@ -0,0 +1,258 @@
+#!/usr/bin/env python3
+"""
+用本地GPS数据(G7日里程 + 东方北斗月里程)替换125辆测试车辆的实际行驶里程。
+测试里程 = TBOX - GPS（负数为0）。
+"""
+import openpyxl, os, shutil, xlrd
+from datetime import datetime, date, timedelta
+from openpyxl.styles import PatternFill, Font, Alignment, Border, Side
+from openpyxl.utils import get_column_letter
+
+os.chdir('/Users/kkfluous/Downloads')
+
+SRC = '租赁任务考核_2026年6月 (1).xlsx'
+DST = '租赁任务考核_2026年6月.xlsx'
+GPS_DIR = 'gps里程汇总2026年1-6月 3'
+
+# ========== 读取125辆测试车辆清单 ==========
+TEST_FILE = '/Users/kkfluous/Library/Containers/com.tencent.xinWeChat/Data/Documents/xwechat_files/wxid_1704407055112_3af8/temp/RWTemp/2026-07/65b41c10aa83cabb7a365b2b8437c1c7/6月测试车辆.xlsx'
+wb_test = openpyxl.load_workbook(TEST_FILE, data_only=True)
+ws_test = wb_test.active
+TEST_PLATES = set()
+for row in ws_test.iter_rows(min_row=2, values_only=True):
+    for ci in range(len(row)):
+        p = str(row[ci]).strip() if row[ci] else ''
+        if p and p != 'None' and len(p) >= 7:
+            TEST_PLATES.add(p)
+wb_test.close()
+print(f"测试车辆清单: {len(TEST_PLATES)}辆")
+
+# ========== Step 1: 加载G7 GPS日里程 ==========
+print("\n加载G7 GPS日里程...")
+g7_wb = openpyxl.load_workbook(f'{GPS_DIR}/G7/202606_G7.xlsx')
+g7_ws = g7_wb.active
+gps_daily = {}   # plate -> [day1, day2, ..., day30]
+for row in g7_ws.iter_rows(min_row=2, values_only=True):
+    plate = str(row[0]).strip() if row[0] else ''
+    if not plate or plate not in TEST_PLATES:
+        continue
+    daily = [float(row[c]) if row[c] else 0 for c in range(3, 33)]
+    gps_daily[plate] = daily
+g7_wb.close()
+print(f"G7覆盖: {len(gps_daily)}辆测试车")
+
+# ========== Step 2: 加载东方北斗GPS月里程 ==========
+print("加载东方北斗GPS月里程...")
+df_wb = openpyxl.load_workbook(f'{GPS_DIR}/东方北斗/202606_东方北斗.xlsx')
+df_ws = df_wb.active
+gps_monthly = {}  # plate -> monthly_total_km
+for row in df_ws.iter_rows(min_row=4, values_only=True):
+    plate = str(row[5]).strip() if row[5] else ''
+    if not plate or plate not in TEST_PLATES:
+        continue
+    monthly = float(row[10]) if row[10] else 0
+    gps_monthly[plate] = monthly
+df_wb.close()
+print(f"东方北斗覆盖: {len(gps_monthly)}辆测试车")
+
+# ========== Step 3: 读取源文件，执行GPS替换 ==========
+print("\n执行GPS替换...")
+
+# 备份
+BAK = '租赁任务考核_2026年6月_备份_0715.xlsx'
+if os.path.exists(DST):
+    shutil.copy2(DST, BAK)
+
+shutil.copy2(SRC, DST)
+
+wb = openpyxl.load_workbook(DST)
+ws = wb['业务考核视图']
+h = [c.value for c in next(ws.iter_rows(min_row=1, max_row=1))]
+print(f"源文件: {ws.max_row}行, {len(h)}列")
+
+# 读取所有记录
+recs = []
+for row in ws.iter_rows(min_row=2, values_only=True):
+    r = dict(zip(h, row))
+    r['row_num'] = len(recs) + 2
+    recs.append(r)
+
+test_recs = [r for r in recs if r['车牌号'] in TEST_PLATES]
+print(f"测试车辆记录: {len(test_recs)}条 ({len(set(r['车牌号'] for r in test_recs))}辆)")
+
+# 添加列
+if '原TBOX里程(km)' not in h:
+    tbox_new_col = ws.max_column + 1
+    test_new_col = ws.max_column + 2
+    ws.cell(row=1, column=tbox_new_col, value='原TBOX里程(km)')
+    ws.cell(row=1, column=test_new_col, value='测试里程(km)')
+    h_new = [c.value for c in next(ws.iter_rows(min_row=1, max_row=1))]
+else:
+    h_new = h
+
+actual_idx = h_new.index('实际行驶里程(km)') + 1
+tbox_idx = h_new.index('原TBOX里程(km)') + 1
+test_idx = h_new.index('测试里程(km)') + 1
+rate_idx = h_new.index('完成率(%)') + 1
+qualified_idx = h_new.index('是否达标') + 1
+
+yellow_fill = PatternFill(start_color='FFFF00', end_color='FFFF00', fill_type='solid')
+
+replaced_g7 = 0
+replaced_df = 0
+no_gps = 0
+total_test_km = 0
+total_gps_km = 0
+total_tbox_km = 0
+
+for r in test_recs:
+    plate = r['车牌号']
+    rn = r['row_num']
+
+    # 获取考核区间
+    start_date = r.get('考核开始日期')
+    end_date = r.get('考核结束日期')
+
+    gps_total = None
+    source = None
+
+    # 优先用G7日数据
+    if plate in gps_daily:
+        daily = gps_daily[plate]
+        try:
+            sd = start_date.date() if hasattr(start_date, 'date') else datetime.strptime(str(start_date)[:10], '%Y-%m-%d').date()
+            ed = end_date.date() if hasattr(end_date, 'date') else datetime.strptime(str(end_date)[:10], '%Y-%m-%d').date()
+        except:
+            sd = date(2026, 6, 1)
+            ed = date(2026, 6, 30)
+
+        gps_total = 0
+        for day_offset in range(30):
+            day = date(2026, 6, 1) + timedelta(days=day_offset)
+            if sd <= day <= ed:
+                gps_total += daily[day_offset]
+        gps_total = round(gps_total, 1)
+        source = 'G7'
+        replaced_g7 += 1
+
+    # 其次用东方北斗月里程
+    elif plate in gps_monthly:
+        # 东方北斗只有月度合计，考核区间一般都是整月
+        gps_total = gps_monthly[plate]
+        source = '东方北斗'
+        replaced_df += 1
+
+    if gps_total is None:
+        no_gps += 1
+        continue
+
+    # 原TBOX里程 = 当前实际行驶里程
+    tbox_val = float(ws.cell(row=rn, column=actual_idx).value or 0)
+
+    # 测试里程 = TBOX - GPS（负数为0）
+    test_km = max(0, round(tbox_val - gps_total, 1))
+
+    # 写入
+    ws.cell(row=rn, column=tbox_idx, value=tbox_val)
+    ws.cell(row=rn, column=test_idx, value=test_km)
+    ws.cell(row=rn, column=actual_idx, value=gps_total)
+
+    # 标记黄色
+    for c in range(1, ws.max_column + 1):
+        ws.cell(row=rn, column=c).fill = yellow_fill
+
+    # 重算完成率
+    target = float(r.get('应考核里程(km)') or 0)
+    pct = round(gps_total / target * 100, 2) if target > 0 else 0
+    qualified = '达标' if gps_total >= target and target > 0 else '未达标'
+    ws.cell(row=rn, column=rate_idx, value=pct)
+    ws.cell(row=rn, column=qualified_idx, value=qualified)
+
+    total_test_km += test_km
+    total_gps_km += gps_total
+    total_tbox_km += tbox_val
+
+print(f"\n替换统计:")
+print(f"  G7日数据: {replaced_g7}条")
+print(f"  东方北斗月数据: {replaced_df}条")
+print(f"  无GPS数据: {no_gps}条")
+print(f"  TBOX总里程: {total_tbox_km:.1f}")
+print(f"  GPS总里程: {total_gps_km:.1f}")
+print(f"  测试总里程: {total_test_km:.1f}")
+
+# ========== Step 4: TBOX回填（非测试车辆） ==========
+print("\nTBOX回填（非测试车辆）...")
+filled = 0
+for row_idx in range(2, ws.max_row + 1):
+    tbox_val = ws.cell(row=row_idx, column=tbox_idx).value
+    actual_val = ws.cell(row=row_idx, column=actual_idx).value
+    test_val = ws.cell(row=row_idx, column=test_idx).value
+    if (tbox_val is None or float(tbox_val) == 0):
+        if test_val is None or float(test_val) == 0:
+            ws.cell(row=row_idx, column=tbox_idx, value=float(actual_val) if actual_val else 0)
+            filled += 1
+print(f"TBOX回填: {filled}条")
+
+# ========== Step 5: 列序排版 ==========
+print("\n列序排版...")
+target_order = [
+    '车牌号', '考核年份', '考核月份', '部门名称', '销售经理', '客户名称',
+    '合同编号', '项目名称', '考核目标ID', '考核目标', '考核开始日期', '考核结束日期',
+    '考核天数', '应考核里程(km)', '原TBOX里程(km)', '实际行驶里程(km)', '测试里程(km)',
+    '完成率(%)', '是否达标', '考核状态', '交车日期', '预计还车日期', '实际还车日期'
+]
+
+h_cur = [c.value for c in next(ws.iter_rows(min_row=1, max_row=1))]
+all_rows = []
+for row_idx in range(1, ws.max_row + 1):
+    row_data = {}
+    for c in range(1, ws.max_column + 1):
+        row_data[h_cur[c-1]] = ws.cell(row=row_idx, column=c).value
+    all_rows.append(row_data)
+
+# Clear
+for r in range(1, ws.max_row + 1):
+    for c in range(1, ws.max_column + 1):
+        ws.cell(row=r, column=c, value=None)
+
+thin_border = Border(left=Side(style='thin'), right=Side(style='thin'),
+                     top=Side(style='thin'), bottom=Side(style='thin'))
+header_fill = PatternFill(start_color='4472C4', end_color='4472C4', fill_type='solid')
+header_font = Font(name='宋体', size=10, bold=True, color='FFFFFF')
+data_font = Font(name='宋体', size=10)
+
+for i, col_name in enumerate(target_order):
+    c = i + 1
+    cell = ws.cell(row=1, column=c, value=col_name)
+    cell.font = header_font
+    cell.fill = header_fill
+    cell.alignment = Alignment(horizontal='center', vertical='center')
+    cell.border = thin_border
+
+for row_idx, row_data in enumerate(all_rows[1:], start=2):
+    for i, col_name in enumerate(target_order):
+        c = i + 1
+        val = row_data.get(col_name)
+        cell = ws.cell(row=row_idx, column=c, value=val)
+        cell.font = data_font
+        cell.alignment = Alignment(vertical='center')
+        cell.border = thin_border
+        if col_name in ['考核开始日期', '考核结束日期', '交车日期', '预计还车日期', '实际还车日期']:
+            cell.number_format = 'yyyy-mm-dd h:mm:ss'
+        if row_data.get('测试里程(km)') and float(row_data.get('测试里程(km)', 0) or 0) != 0:
+            cell.fill = yellow_fill
+
+for i in range(len(target_order)):
+    ws.column_dimensions[get_column_letter(i+1)].width = 15
+
+wb.save(DST)
+wb.close()
+
+print(f"列序排版: {len(target_order)}列")
+print(f"\n{'='*60}")
+print(f"GPS替换完成！")
+print(f"测试车辆: {len(TEST_PLATES)}辆 -> 源数据{len(test_recs)}条记录")
+print(f"GPS来源: G7({replaced_g7}条) + 东方北斗({replaced_df}条)")
+print(f"测试总里程: {total_test_km:.1f} km")
+print(f"已保存: {DST}")
+print(f"{'='*60}")
diff --git a/gps_replace_local.py b/gps_replace_local.py
new file mode 100644
index 0000000..0a5efcb
--- /dev/null
+++ b/gps_replace_local.py
@@ -0,0 +1,264 @@
+#!/usr/bin/env python3
+"""
+从本地GPS数据(G7)整合6月GPS里程，替换测试车辆实际行驶里程
+测试里程 = TBOX总里程 - GPS总里程（负数为0）
+重新核算6月里程考核，输出到 ./20260713_1500/
+"""
+import openpyxl, os, shutil
+from datetime import datetime, date
+from openpyxl.styles import PatternFill, Font, Alignment, Border, Side
+from openpyxl.utils import get_column_letter
+
+os.chdir('/Users/kkfluous/Downloads')
+
+SRC = '租赁任务考核_2026年6月 (1).xlsx'
+DST = '租赁任务考核_2026年6月.xlsx'
+GPS_DIR = 'gps里程汇总2026年1-6月 3'
+
+# 36辆测试车辆
+TEST_PLATES = {
+    '粤AGE5480','粤AGF4535','粤AGG3490','粤AGG4135','粤AGG4191',
+    '粤AGP2033','粤AGP3080','粤AGP3663','粤AGP3676','粤AGP5139',
+    '粤AGP5156','粤AGP5163','粤AGP5168','粤AGP5636','粤AGP5767',
+    '粤AGP5768','粤AGP5796','粤AGP9733','粤AGP9786','粤AGR0298',
+    '粤AGR1586','粤AGR3288','粤AGR5056','粤AGR5068','粤AGR5088',
+    '粤AGR6869','粤AGR6877','粤AGR6879','粤AGR6881','粤AGR8506',
+    '粤AGR8551','粤AGR8558','粤AGR8799','粤AGR9866','粤AGR9893',
+    '粤AGR9899',
+}
+
+# ========== Step 1: 加载G7 GPS日里程数据 ==========
+print("=" * 60)
+print("Step 1: 加载G7 GPS日里程数据")
+print("=" * 60)
+
+g7_wb = openpyxl.load_workbook(f'{GPS_DIR}/G7/202606_G7.xlsx')
+g7_ws = g7_wb.active
+
+# G7格式: 车牌号(0), 机构(1), 行驶里程(2), 06月01日(3)...06月30日(32), 运行时长(33), 数据完整度说明(34)
+gps_daily = {}  # plate -> [day1_km, day2_km, ..., day30_km]
+gps_monthly = {}  # plate -> monthly_total_km
+no_data_plates = set()
+
+for row in g7_ws.iter_rows(min_row=2, values_only=True):
+    plate = str(row[0]).strip() if row[0] else ''
+    if not plate:
+        continue
+    monthly_km = float(row[2]) if row[2] else 0
+    daily = []
+    for c in range(3, 33):  # 06月01日 to 06月30日
+        daily.append(float(row[c]) if row[c] else 0)
+    gps_daily[plate] = daily
+    gps_monthly[plate] = monthly_km
+
+g7_wb.close()
+print(f"G7加载完成: {len(gps_daily)}辆车")
+
+# ========== Step 2: 读取源文件，执行GPS替换 ==========
+print("\n" + "=" * 60)
+print("Step 2: GPS里程替换")
+print("=" * 60)
+
+# 备份现有文件
+BAK = '租赁任务考核_2026年6月_备份_0715.xlsx'
+if os.path.exists(DST):
+    shutil.copy2(DST, BAK)
+    print(f'已备份: {BAK}')
+
+# 从原始新文件开始
+shutil.copy2(SRC, DST)
+print(f'源文件: {SRC} → {DST}')
+
+wb = openpyxl.load_workbook(DST)
+ws = wb['业务考核视图']
+h = [c.value for c in next(ws.iter_rows(min_row=1, max_row=1))]
+print(f'列数: {len(h)}')
+
+# 读取所有记录
+recs = []
+for row in ws.iter_rows(min_row=2, values_only=True):
+    r = dict(zip(h, row))
+    r['row_num'] = len(recs) + 2
+    recs.append(r)
+
+# 找测试车辆记录
+test_recs = [r for r in recs if r['车牌号'] in TEST_PLATES]
+print(f'测试车辆记录: {len(test_recs)}条 ({len(set(r["车牌号"] for r in test_recs))}辆)')
+
+# 添加 原TBOX里程(km) 和 测试里程(km) 列
+if '原TBOX里程(km)' not in h:
+    tbox_new_col = ws.max_column + 1
+    test_new_col = ws.max_column + 2
+    ws.cell(row=1, column=tbox_new_col, value='原TBOX里程(km)')
+    ws.cell(row=1, column=test_new_col, value='测试里程(km)')
+    h_new = [c.value for c in next(ws.iter_rows(min_row=1, max_row=1))]
+else:
+    h_new = h
+
+actual_idx = h_new.index('实际行驶里程(km)') + 1
+tbox_idx = h_new.index('原TBOX里程(km)') + 1
+test_idx = h_new.index('测试里程(km)') + 1
+rate_idx = h_new.index('完成率(%)') + 1
+qualified_idx = h_new.index('是否达标') + 1
+
+yellow_fill = PatternFill(start_color='FFFF00', end_color='FFFF00', fill_type='solid')
+
+DAYSM = {1: 31, 2: 28, 3: 31, 4: 30, 5: 31, 6: 30}
+
+replaced = 0
+no_gps = 0
+no_gps_plates = set()
+
+for r in test_recs:
+    plate = r['车牌号']
+    rn = r['row_num']
+
+    # 获取考核起止日期
+    start_date = r.get('考核开始日期')
+    end_date = r.get('考核结束日期')
+
+    if plate not in gps_daily:
+        no_gps += 1
+        no_gps_plates.add(plate)
+        continue
+
+    daily = gps_daily[plate]
+
+    # 计算考核区间内的GPS总里程
+    if start_date and end_date:
+        try:
+            sd = start_date.date() if hasattr(start_date, 'date') else datetime.strptime(str(start_date)[:10], '%Y-%m-%d').date()
+            ed = end_date.date() if hasattr(end_date, 'date') else datetime.strptime(str(end_date)[:10], '%Y-%m-%d').date()
+        except:
+            sd = date(2026, 6, 1)
+            ed = date(2026, 6, 30)
+    else:
+        sd = date(2026, 6, 1)
+        ed = date(2026, 6, 30)
+
+    # 汇总考核区间内的GPS日里程（G7日数据索引: 0=6/1, 29=6/30）
+    gps_total = 0
+    for day_offset in range(30):
+        day = date(2026, 6, 1) + date.resolution * day_offset  # timedelta
+        from datetime import timedelta
+        day = date(2026, 6, 1) + timedelta(days=day_offset)
+        if sd <= day <= ed:
+            gps_total += daily[day_offset]
+
+    gps_total = round(gps_total, 1)
+
+    # 原TBOX里程 = 当前实际行驶里程
+    tbox_val = float(ws.cell(row=rn, column=actual_idx).value or 0)
+
+    # 测试里程 = TBOX - GPS（负数为0）
+    test_km = max(0, round(tbox_val - gps_total, 1))
+
+    # 写入
+    ws.cell(row=rn, column=tbox_idx, value=tbox_val)
+    ws.cell(row=rn, column=test_idx, value=test_km)
+    ws.cell(row=rn, column=actual_idx, value=gps_total)
+
+    # 标记黄色
+    for c in range(1, ws.max_column + 1):
+        ws.cell(row=rn, column=c).fill = yellow_fill
+
+    # 重算完成率
+    target = float(r.get('应考核里程(km)') or 0)
+    pct = round(gps_total / target * 100, 2) if target > 0 else 0
+    qualified = '达标' if gps_total >= target and target > 0 else '未达标'
+    ws.cell(row=rn, column=rate_idx, value=pct)
+    ws.cell(row=rn, column=qualified_idx, value=qualified)
+
+    replaced += 1
+    print(f"  {plate}: TBOX={tbox_val}, GPS={gps_total}, 测试={test_km}, 达标={qualified}")
+
+print(f'\nGPS替换: 成功{replaced}条, 无GPS数据{no_gps}条')
+if no_gps_plates:
+    print(f'无GPS数据车牌: {sorted(no_gps_plates)}')
+
+# ========== Step 3: TBOX回填（非测试车辆） ==========
+print("\n" + "=" * 60)
+print("Step 3: TBOX回填（非测试车辆）")
+print("=" * 60)
+
+filled = 0
+for row_idx in range(2, ws.max_row + 1):
+    tbox_val = ws.cell(row=row_idx, column=tbox_idx).value
+    actual_val = ws.cell(row=row_idx, column=actual_idx).value
+    test_val = ws.cell(row=row_idx, column=test_idx).value
+
+    if (tbox_val is None or float(tbox_val) == 0):
+        if test_val is None or float(test_val) == 0:
+            ws.cell(row=row_idx, column=tbox_idx, value=float(actual_val) if actual_val else 0)
+            filled += 1
+
+print(f'TBOX回填: {filled}条')
+
+# ========== Step 4: 列序排版 ==========
+print("\n" + "=" * 60)
+print("Step 4: 列序排版")
+print("=" * 60)
+
+target_order = [
+    '车牌号', '考核年份', '考核月份', '部门名称', '销售经理', '客户名称',
+    '合同编号', '项目名称', '考核目标ID', '考核目标', '考核开始日期', '考核结束日期',
+    '考核天数', '应考核里程(km)', '原TBOX里程(km)', '实际行驶里程(km)', '测试里程(km)',
+    '完成率(%)', '是否达标', '考核状态', '交车日期', '预计还车日期', '实际还车日期'
+]
+
+# Read all data with styles
+h_cur = [c.value for c in next(ws.iter_rows(min_row=1, max_row=1))]
+all_rows = []
+for row_idx in range(1, ws.max_row + 1):
+    row_data = {}
+    for c in range(1, ws.max_column + 1):
+        row_data[h_cur[c-1]] = ws.cell(row=row_idx, column=c).value
+    all_rows.append(row_data)
+
+# Clear sheet
+for r in range(1, ws.max_row + 1):
+    for c in range(1, ws.max_column + 1):
+        ws.cell(row=r, column=c, value=None)
+
+# Apply styles
+thin_border = Border(left=Side(style='thin'), right=Side(style='thin'),
+                     top=Side(style='thin'), bottom=Side(style='thin'))
+header_fill = PatternFill(start_color='4472C4', end_color='4472C4', fill_type='solid')
+header_font = Font(name='宋体', size=10, bold=True, color='FFFFFF')
+data_font = Font(name='宋体', size=10)
+
+# Write headers
+for i, col_name in enumerate(target_order):
+    c = i + 1
+    cell = ws.cell(row=1, column=c, value=col_name)
+    cell.font = header_font
+    cell.fill = header_fill
+    cell.alignment = Alignment(horizontal='center', vertical='center')
+    cell.border = thin_border
+
+# Write data
+for row_idx, row_data in enumerate(all_rows[1:], start=2):
+    for i, col_name in enumerate(target_order):
+        c = i + 1
+        val = row_data.get(col_name)
+        cell = ws.cell(row=row_idx, column=c, value=val)
+        cell.font = data_font
+        cell.alignment = Alignment(vertical='center')
+        cell.border = thin_border
+        if col_name in ['考核开始日期', '考核结束日期', '交车日期', '预计还车日期', '实际还车日期']:
+            cell.number_format = 'yyyy-mm-dd h:mm:ss'
+        # Restore yellow fill for test vehicles
+        if row_data.get('测试里程(km)') and float(row_data.get('测试里程(km)', 0) or 0) != 0:
+            cell.fill = yellow_fill
+
+for i in range(len(target_order)):
+    ws.column_dimensions[get_column_letter(i+1)].width = 15
+
+wb.save(DST)
+wb.close()
+print(f'列序排版完成 ({len(target_order)}列)')
+print(f'已保存: {DST}')
+
+print("\n" + "=" * 60)
+print("GPS替换完成！接下来运行 main.py...")
+print("=" * 60)
diff --git a/index (1).html b/index (1).html
new file mode 100644
index 0000000..5ef77db
--- /dev/null
+++ b/index (1).html	
@@ -0,0 +1,107 @@
+<!doctype html>
+<html>
+<head>
+  <meta charset="utf-8">
+  <title>小羚羚小程序</title>
+  <meta name="viewport" content="width=device-width,initial-scale=1">
+  <style>
+    html, body {
+      box-sizing: border-box;
+      width: 100%;
+      margin: 0;
+      padding: 0;
+      min-height: 100%;
+      height: 100%;
+      overflow-x: hidden;
+      overflow-y: auto;
+    }
+    #root {
+      width: 100%;
+      margin-left: auto;
+      margin-right: auto;
+      height: 100%;
+      min-height: 100vh;
+      overflow: visible;
+    }
+    body.is-element-page #root {
+      width: 100vw;
+      height: 100vh;
+    }
+  </style>
+</head>
+<body>
+
+  <script>
+    (function () {
+      var isComponent = false;
+      if (document.body && (isComponent || window.location.pathname.includes('/components/'))) {
+        document.body.classList.add('is-element-page');
+      }
+    })();
+  </script>
+
+  <div id="root"></div>
+
+  <script>window.__AXHUB_ANNOTATION_SOURCE_REFERENCE__={root:"source",manifest:"source/manifest.json"};</script>
+
+  <script src="./assets/react.production.min.js"></script>
+  <script src="./assets/react-dom.production.min.js"></script>
+  <script src="./assets/export-html-bootstrap.js"></script>
+
+  <script>
+    function waitForBootstrap(timeoutMs) {
+      timeoutMs = timeoutMs || 10000;
+      return new Promise(function (resolve, reject) {
+        var startedAt = Date.now();
+        function check() {
+          if (window.HtmlTemplateBootstrap) {
+            resolve(window.HtmlTemplateBootstrap);
+            return;
+          }
+          if (Date.now() - startedAt >= timeoutMs) {
+            reject(new Error('[Html Template] Bootstrap 初始化超时'));
+            return;
+          }
+          setTimeout(check, 10);
+        }
+        check();
+      });
+    }
+
+    function loadEntryScript(src) {
+      return new Promise(function (resolve, reject) {
+        var script = document.createElement('script');
+        script.src = src;
+        script.async = false;
+        script.onload = function () { resolve(); };
+        script.onerror = function () { reject(new Error('[Html Template] 入口脚本加载失败: ' + src)); };
+        document.body.appendChild(script);
+      });
+    }
+
+    async function bootstrapAndRender() {
+      try {
+        var bootstrap = await waitForBootstrap();
+        var renderComponent = bootstrap.renderComponent;
+        window.React = bootstrap.React;
+        window.ReactDOM = bootstrap.ReactDOM;
+
+        await loadEntryScript('./index.js');
+        window.__AXHUB_ANNOTATION_SOURCE_REFERENCE__={root:"source",manifest:"source/manifest.json"};
+
+        var Component = window.UserComponent && (window.UserComponent.Component || window.UserComponent.default || window.UserComponent);
+        if (!Component) {
+          throw new Error('[Html Template] 入口脚本未暴露 UserComponent');
+        }
+
+        renderComponent(Component);
+      } catch (error) {
+        console.error('[Html Template] 页面初始化失败:', error);
+      }
+    }
+
+    bootstrapAndRender();
+  </script>
+
+</body>
+</html>
\ No newline at end of file
diff --git a/index (2).html b/index (2).html
new file mode 100644
index 0000000..5ef77db
--- /dev/null
+++ b/index (2).html	
@@ -0,0 +1,107 @@
+<!doctype html>
+<html>
+<head>
+  <meta charset="utf-8">
+  <title>小羚羚小程序</title>
+  <meta name="viewport" content="width=device-width,initial-scale=1">
+  <style>
+    html, body {
+      box-sizing: border-box;
+      width: 100%;
+      margin: 0;
+      padding: 0;
+      min-height: 100%;
+      height: 100%;
+      overflow-x: hidden;
+      overflow-y: auto;
+    }
+    #root {
+      width: 100%;
+      margin-left: auto;
+      margin-right: auto;
+      height: 100%;
+      min-height: 100vh;
+      overflow: visible;
+    }
+    body.is-element-page #root {
+      width: 100vw;
+      height: 100vh;
+    }
+  </style>
+</head>
+<body>
+
+  <script>
+    (function () {
+      var isComponent = false;
+      if (document.body && (isComponent || window.location.pathname.includes('/components/'))) {
+        document.body.classList.add('is-element-page');
+      }
+    })();
+  </script>
+
+  <div id="root"></div>
+
+  <script>window.__AXHUB_ANNOTATION_SOURCE_REFERENCE__={root:"source",manifest:"source/manifest.json"};</script>
+
+  <script src="./assets/react.production.min.js"></script>
+  <script src="./assets/react-dom.production.min.js"></script>
+  <script src="./assets/export-html-bootstrap.js"></script>
+
+  <script>
+    function waitForBootstrap(timeoutMs) {
+      timeoutMs = timeoutMs || 10000;
+      return new Promise(function (resolve, reject) {
+        var startedAt = Date.now();
+        function check() {
+          if (window.HtmlTemplateBootstrap) {
+            resolve(window.HtmlTemplateBootstrap);
+            return;
+          }
+          if (Date.now() - startedAt >= timeoutMs) {
+            reject(new Error('[Html Template] Bootstrap 初始化超时'));
+            return;
+          }
+          setTimeout(check, 10);
+        }
+        check();
+      });
+    }
+
+    function loadEntryScript(src) {
+      return new Promise(function (resolve, reject) {
+        var script = document.createElement('script');
+        script.src = src;
+        script.async = false;
+        script.onload = function () { resolve(); };
+        script.onerror = function () { reject(new Error('[Html Template] 入口脚本加载失败: ' + src)); };
+        document.body.appendChild(script);
+      });
+    }
+
+    async function bootstrapAndRender() {
+      try {
+        var bootstrap = await waitForBootstrap();
+        var renderComponent = bootstrap.renderComponent;
+        window.React = bootstrap.React;
+        window.ReactDOM = bootstrap.ReactDOM;
+
+        await loadEntryScript('./index.js');
+        window.__AXHUB_ANNOTATION_SOURCE_REFERENCE__={root:"source",manifest:"source/manifest.json"};
+
+        var Component = window.UserComponent && (window.UserComponent.Component || window.UserComponent.default || window.UserComponent);
+        if (!Component) {
+          throw new Error('[Html Template] 入口脚本未暴露 UserComponent');
+        }
+
+        renderComponent(Component);
+      } catch (error) {
+        console.error('[Html Template] 页面初始化失败:', error);
+      }
+    }
+
+    bootstrapAndRender();
+  </script>
+
+</body>
+</html>
\ No newline at end of file
diff --git a/index (3).html b/index (3).html
new file mode 100644
index 0000000..5ef77db
--- /dev/null
+++ b/index (3).html	
@@ -0,0 +1,107 @@
+<!doctype html>
+<html>
+<head>
+  <meta charset="utf-8">
+  <title>小羚羚小程序</title>
+  <meta name="viewport" content="width=device-width,initial-scale=1">
+  <style>
+    html, body {
+      box-sizing: border-box;
+      width: 100%;
+      margin: 0;
+      padding: 0;
+      min-height: 100%;
+      height: 100%;
+      overflow-x: hidden;
+      overflow-y: auto;
+    }
+    #root {
+      width: 100%;
+      margin-left: auto;
+      margin-right: auto;
+      height: 100%;
+      min-height: 100vh;
+      overflow: visible;
+    }
+    body.is-element-page #root {
+      width: 100vw;
+      height: 100vh;
+    }
+  </style>
+</head>
+<body>
+
+  <script>
+    (function () {
+      var isComponent = false;
+      if (document.body && (isComponent || window.location.pathname.includes('/components/'))) {
+        document.body.classList.add('is-element-page');
+      }
+    })();
+  </script>
+
+  <div id="root"></div>
+
+  <script>window.__AXHUB_ANNOTATION_SOURCE_REFERENCE__={root:"source",manifest:"source/manifest.json"};</script>
+
+  <script src="./assets/react.production.min.js"></script>
+  <script src="./assets/react-dom.production.min.js"></script>
+  <script src="./assets/export-html-bootstrap.js"></script>
+
+  <script>
+    function waitForBootstrap(timeoutMs) {
+      timeoutMs = timeoutMs || 10000;
+      return new Promise(function (resolve, reject) {
+        var startedAt = Date.now();
+        function check() {
+          if (window.HtmlTemplateBootstrap) {
+            resolve(window.HtmlTemplateBootstrap);
+            return;
+          }
+          if (Date.now() - startedAt >= timeoutMs) {
+            reject(new Error('[Html Template] Bootstrap 初始化超时'));
+            return;
+          }
+          setTimeout(check, 10);
+        }
+        check();
+      });
+    }
+
+    function loadEntryScript(src) {
+      return new Promise(function (resolve, reject) {
+        var script = document.createElement('script');
+        script.src = src;
+        script.async = false;
+        script.onload = function () { resolve(); };
+        script.onerror = function () { reject(new Error('[Html Template] 入口脚本加载失败: ' + src)); };
+        document.body.appendChild(script);
+      });
+    }
+
+    async function bootstrapAndRender() {
+      try {
+        var bootstrap = await waitForBootstrap();
+        var renderComponent = bootstrap.renderComponent;
+        window.React = bootstrap.React;
+        window.ReactDOM = bootstrap.ReactDOM;
+
+        await loadEntryScript('./index.js');
+        window.__AXHUB_ANNOTATION_SOURCE_REFERENCE__={root:"source",manifest:"source/manifest.json"};
+
+        var Component = window.UserComponent && (window.UserComponent.Component || window.UserComponent.default || window.UserComponent);
+        if (!Component) {
+          throw new Error('[Html Template] 入口脚本未暴露 UserComponent');
+        }
+
+        renderComponent(Component);
+      } catch (error) {
+        console.error('[Html Template] 页面初始化失败:', error);
+      }
+    }
+
+    bootstrapAndRender();
+  </script>
+
+</body>
+</html>
\ No newline at end of file
diff --git a/index (4).html b/index (4).html
new file mode 100644
index 0000000..5ef77db
--- /dev/null
+++ b/index (4).html	
@@ -0,0 +1,107 @@
+<!doctype html>
+<html>
+<head>
+  <meta charset="utf-8">
+  <title>小羚羚小程序</title>
+  <meta name="viewport" content="width=device-width,initial-scale=1">
+  <style>
+    html, body {
+      box-sizing: border-box;
+      width: 100%;
+      margin: 0;
+      padding: 0;
+      min-height: 100%;
+      height: 100%;
+      overflow-x: hidden;
+      overflow-y: auto;
+    }
+    #root {
+      width: 100%;
+      margin-left: auto;
+      margin-right: auto;
+      height: 100%;
+      min-height: 100vh;
+      overflow: visible;
+    }
+    body.is-element-page #root {
+      width: 100vw;
+      height: 100vh;
+    }
+  </style>
+</head>
+<body>
+
+  <script>
+    (function () {
+      var isComponent = false;
+      if (document.body && (isComponent || window.location.pathname.includes('/components/'))) {
+        document.body.classList.add('is-element-page');
+      }
+    })();
+  </script>
+
+  <div id="root"></div>
+
+  <script>window.__AXHUB_ANNOTATION_SOURCE_REFERENCE__={root:"source",manifest:"source/manifest.json"};</script>
+
+  <script src="./assets/react.production.min.js"></script>
+  <script src="./assets/react-dom.production.min.js"></script>
+  <script src="./assets/export-html-bootstrap.js"></script>
+
+  <script>
+    function waitForBootstrap(timeoutMs) {
+      timeoutMs = timeoutMs || 10000;
+      return new Promise(function (resolve, reject) {
+        var startedAt = Date.now();
+        function check() {
+          if (window.HtmlTemplateBootstrap) {
+            resolve(window.HtmlTemplateBootstrap);
+            return;
+          }
+          if (Date.now() - startedAt >= timeoutMs) {
+            reject(new Error('[Html Template] Bootstrap 初始化超时'));
+            return;
+          }
+          setTimeout(check, 10);
+        }
+        check();
+      });
+    }
+
+    function loadEntryScript(src) {
+      return new Promise(function (resolve, reject) {
+        var script = document.createElement('script');
+        script.src = src;
+        script.async = false;
+        script.onload = function () { resolve(); };
+        script.onerror = function () { reject(new Error('[Html Template] 入口脚本加载失败: ' + src)); };
+        document.body.appendChild(script);
+      });
+    }
+
+    async function bootstrapAndRender() {
+      try {
+        var bootstrap = await waitForBootstrap();
+        var renderComponent = bootstrap.renderComponent;
+        window.React = bootstrap.React;
+        window.ReactDOM = bootstrap.ReactDOM;
+
+        await loadEntryScript('./index.js');
+        window.__AXHUB_ANNOTATION_SOURCE_REFERENCE__={root:"source",manifest:"source/manifest.json"};
+
+        var Component = window.UserComponent && (window.UserComponent.Component || window.UserComponent.default || window.UserComponent);
+        if (!Component) {
+          throw new Error('[Html Template] 入口脚本未暴露 UserComponent');
+        }
+
+        renderComponent(Component);
+      } catch (error) {
+        console.error('[Html Template] 页面初始化失败:', error);
+      }
+    }
+
+    bootstrapAndRender();
+  </script>
+
+</body>
+</html>
\ No newline at end of file
diff --git a/index.html b/index.html
new file mode 100644
index 0000000..5ef77db
--- /dev/null
+++ b/index.html
@@ -0,0 +1,107 @@
+<!doctype html>
+<html>
+<head>
+  <meta charset="utf-8">
+  <title>小羚羚小程序</title>
+  <meta name="viewport" content="width=device-width,initial-scale=1">
+  <style>
+    html, body {
+      box-sizing: border-box;
+      width: 100%;
+      margin: 0;
+      padding: 0;
+      min-height: 100%;
+      height: 100%;
+      overflow-x: hidden;
+      overflow-y: auto;
+    }
+    #root {
+      width: 100%;
+      margin-left: auto;
+      margin-right: auto;
+      height: 100%;
+      min-height: 100vh;
+      overflow: visible;
+    }
+    body.is-element-page #root {
+      width: 100vw;
+      height: 100vh;
+    }
+  </style>
+</head>
+<body>
+
+  <script>
+    (function () {
+      var isComponent = false;
+      if (document.body && (isComponent || window.location.pathname.includes('/components/'))) {
+        document.body.classList.add('is-element-page');
+      }
+    })();
+  </script>
+
+  <div id="root"></div>
+
+  <script>window.__AXHUB_ANNOTATION_SOURCE_REFERENCE__={root:"source",manifest:"source/manifest.json"};</script>
+
+  <script src="./assets/react.production.min.js"></script>
+  <script src="./assets/react-dom.production.min.js"></script>
+  <script src="./assets/export-html-bootstrap.js"></script>
+
+  <script>
+    function waitForBootstrap(timeoutMs) {
+      timeoutMs = timeoutMs || 10000;
+      return new Promise(function (resolve, reject) {
+        var startedAt = Date.now();
+        function check() {
+          if (window.HtmlTemplateBootstrap) {
+            resolve(window.HtmlTemplateBootstrap);
+            return;
+          }
+          if (Date.now() - startedAt >= timeoutMs) {
+            reject(new Error('[Html Template] Bootstrap 初始化超时'));
+            return;
+          }
+          setTimeout(check, 10);
+        }
+        check();
+      });
+    }
+
+    function loadEntryScript(src) {
+      return new Promise(function (resolve, reject) {
+        var script = document.createElement('script');
+        script.src = src;
+        script.async = false;
+        script.onload = function () { resolve(); };
+        script.onerror = function () { reject(new Error('[Html Template] 入口脚本加载失败: ' + src)); };
+        document.body.appendChild(script);
+      });
+    }
+
+    async function bootstrapAndRender() {
+      try {
+        var bootstrap = await waitForBootstrap();
+        var renderComponent = bootstrap.renderComponent;
+        window.React = bootstrap.React;
+        window.ReactDOM = bootstrap.ReactDOM;
+
+        await loadEntryScript('./index.js');
+        window.__AXHUB_ANNOTATION_SOURCE_REFERENCE__={root:"source",manifest:"source/manifest.json"};
+
+        var Component = window.UserComponent && (window.UserComponent.Component || window.UserComponent.default || window.UserComponent);
+        if (!Component) {
+          throw new Error('[Html Template] 入口脚本未暴露 UserComponent');
+        }
+
+        renderComponent(Component);
+      } catch (error) {
+        console.error('[Html Template] 页面初始化失败:', error);
+      }
+    }
+
+    bootstrapAndRender();
+  </script>
+
+</body>
+</html>
\ No newline at end of file
diff --git a/jun_18t_s3.py b/jun_18t_s3.py
new file mode 100644
index 0000000..7cb2d92
--- /dev/null
+++ b/jun_18t_s3.py
@@ -0,0 +1,369 @@
+#!/usr/bin/env python3
+"""
+6月 18T 绩效考核 — 方案3 (800/300 当月, 800结转, 500补发/累计)
+"""
+
+import os, openpyxl, unicodedata
+from collections import defaultdict
+
+os.chdir('/Users/kkfluous/Downloads')
+
+DAYS = {1:31, 2:28, 3:31, 4:30, 5:31, 6:30}
+TARGET_KM = 6000
+BASE_GOOD = 800    # 当月达标
+BASE_BAD = 300     # 当月未达标
+BASE_CARRY = 800   # 结转
+BASE_OTHER = 500   # 补发/累计补发
+OUTPUT_DIR = '/Users/kkfluous/Downloads/6月18T测算'
+
+# ========== 数据读取 ==========
+
+def read_18t(filepath, month):
+    wb = openpyxl.load_workbook(filepath, data_only=True)
+    ws = wb['业务考核视图']
+    h = [c.value for c in next(ws.iter_rows(min_row=1, max_row=1))]
+    recs = []
+    for row in ws.iter_rows(min_row=2, values_only=True):
+        r = dict(zip(h, row))
+        if r.get('考核目标') != '羚牛100辆18T':
+            continue
+        r['考核天数'] = float(r['考核天数'] or 0)
+        r['应考核里程(km)'] = float(r['应考核里程(km)'] or 0)
+        r['实际行驶里程(km)'] = float(r['实际行驶里程(km)'] or 0)
+        r['月份'] = month
+        t, a = r['应考核里程(km)'], r['实际行驶里程(km)']
+        q = a >= t and t > 0
+        r['是否达标'] = '达标' if q else '未达标'
+        r['多跑'] = max(0, a - t) if q else 0
+        r['可结转'] = int(r['多跑'] // TARGET_KM) if q else 0
+        recs.append(r)
+    wb.close()
+    return recs
+
+
+def grp(recs):
+    gs = defaultdict(lambda: {
+        'recs': [], '应考核': 0, '实际': 0, '奖金': 0, '天数': 0,
+        '有达标': False, '部门': '', '销售': '', '车牌': ''
+    })
+    for r in recs:
+        k = (r['车牌号'], r['销售经理'])
+        g = gs[k]
+        g['recs'].append(r)
+        g['应考核'] += r['应考核里程(km)']
+        g['实际'] += r['实际行驶里程(km)']
+        g['奖金'] += r.get('奖金', 0)
+        g['天数'] += r['考核天数']
+        if r['是否达标'] == '达标':
+            g['有达标'] = True
+        g['部门'] = r['部门名称']
+        g['销售'] = r['销售经理']
+        g['车牌'] = r['车牌号']
+    return dict(gs)
+
+
+# ========== 计算 ==========
+
+def calc():
+    D = {}
+    for m in range(1, 7):
+        fp = f'租赁任务考核_2026年{m}月.xlsx'
+        if os.path.exists(fp):
+            D[m] = read_18t(fp, m)
+
+    # 给每条记录算奖金
+    for m in D:
+        for r in D[m]:
+            t, a = r['应考核里程(km)'], r['实际行驶里程(km)']
+            q = a >= t and t > 0
+            r['是否达标'] = '达标' if q else '未达标'
+            if q:
+                r['奖金'] = BASE_GOOD * (r['考核天数'] / DAYS[m])
+            else:
+                r['奖金'] = BASE_BAD * (r['考核天数'] / DAYS[m])
+            r['多跑'] = max(0, a - t) if q else 0
+            r['可结转'] = int(r['多跑'] // TARGET_KM) if q else 0
+
+    G = {m: grp(D[m]) for m in D}
+
+    # 1月: 初始化
+    for k, g in G[1].items():
+        cum_q = g['实际'] >= g['应考核'] and g['应考核'] > 0
+        cum_over = max(0, g['实际'] - g['应考核']) if cum_q else 0
+        g['可结转'] = int(cum_over // TARGET_KM) if TARGET_KM > 0 else 0
+        g['累计已发结转'] = 0
+        g['1月已发'] = g['有达标']
+        g['cum_t'] = g['应考核']
+        g['cum_a'] = g['实际']
+        g['cum_q'] = cum_q
+
+    # 逐月计算 2→6
+    for month in range(2, 7):
+        if month not in G:
+            continue
+        prev_m = month - 1
+        days_m = DAYS[month]
+
+        for k, gm in G[month].items():
+            gp = G[prev_m].get(k)
+
+            cum_t = (gp['cum_t'] if gp else 0) + gm['应考核']
+            cum_a = (gp['cum_a'] if gp else 0) + gm['实际']
+            cum_q = cum_a >= cum_t and cum_t > 0
+
+            # 结转: 800
+            carry = 0
+            p_carry = gp['可结转'] if gp else 0
+            if gp and p_carry >= 1:
+                carry = BASE_CARRY
+
+            # 当月奖金
+            gm['当月奖金'] = gm['奖金'] if carry == 0 else 0
+
+            # 补发历史月: 500
+            for back_m in range(1, month):
+                g_bm = G[back_m].get(k)
+                if not g_bm:
+                    continue
+                bm_paid = g_bm.get(f'{back_m}月已发', g_bm['有达标'])
+                if bm_paid:
+                    continue
+                bm_paid_later = False
+                for later_m in range(back_m + 1, month):
+                    gl = G[later_m].get(k)
+                    if gl and gl.get(f'补发{back_m}月', 0) > 0:
+                        bm_paid_later = True
+                        break
+                if bm_paid_later:
+                    continue
+                if cum_q:
+                    bj = BASE_OTHER * (g_bm['天数'] / DAYS[back_m])
+                    gm[f'补发{back_m}月'] = bj
+
+            # 累计补发当月: 500
+            cbp = 0
+            if not gm['有达标'] and carry == 0 and cum_q:
+                cbp = BASE_OTHER * (gm['天数'] / days_m)
+            gm['累计补发当月'] = cbp
+
+            gm['cum_t'] = cum_t
+            gm['cum_a'] = cum_a
+            gm['cum_q'] = cum_q
+            gm['结转'] = carry
+            gm[f'{month}月已发'] = carry > 0 or gm['有达标'] or cbp > 0
+
+            # 更新可结转
+            cum_over = max(0, cum_a - cum_t) if cum_q else 0
+            earned = int(cum_over // TARGET_KM) if TARGET_KM > 0 else 0
+            prev_used = gp['累计已发结转'] if gp else 0
+            cur_used = prev_used + (1 if carry > 0 else 0)
+            gm['累计已发结转'] = cur_used
+            gm['可结转'] = max(0, earned - cur_used)
+
+        # 补充: 上月有结转但本月无考核记录的车
+        for k, gp in G[prev_m].items():
+            if k not in G[month] and gp.get('可结转', 0) >= 1:
+                G[month][k] = {
+                    'recs': [], '应考核': TARGET_KM, '实际': 0, '奖金': 0, '天数': days_m,
+                    '有达标': False, '部门': gp['部门'], '销售': gp['销售'], '车牌': k[0],
+                    'cum_t': gp.get('cum_t', 0) + TARGET_KM, 'cum_a': gp.get('cum_a', 0),
+                    'cum_q': False,
+                    '结转': BASE_CARRY, '当月奖金': 0,
+                    f'{month}月已发': True,
+                    '可结转': max(0, gp['可结转'] - 1),
+                    '累计已发结转': gp.get('累计已发结转', 0) + 1,
+                    '虚拟': True,
+                }
+
+    # 收集6月发放记录
+    pay_records = collect_payments(G, 6)
+
+    return D[6], G[6], pay_records
+
+
+def collect_payments(G, month):
+    G_m = G[month]
+    records = []
+    for k, g in sorted(G_m.items()):
+        if g.get('虚拟'):
+            continue
+        carry = g.get('结转', 0)
+        if carry > 0:
+            records.append({'车牌': k[0], '销售': k[1], '部门': g['部门'],
+                           '金额': carry, '类型': '结转(800)'})
+
+        bonus = g.get('当月奖金', 0)
+        if bonus > 0:
+            ptype = '当月达标(800)' if g['有达标'] else '当月未达标(300)'
+            records.append({'车牌': k[0], '销售': k[1], '部门': g['部门'],
+                           '金额': bonus, '类型': ptype})
+
+        cbp = g.get('累计补发当月', 0)
+        if cbp > 0:
+            records.append({'车牌': k[0], '销售': k[1], '部门': g['部门'],
+                           '金额': cbp, '类型': f'累计补发{month}月(500)'})
+
+        for bm in range(1, month):
+            bv = g.get(f'补发{bm}月', 0)
+            if bv > 0:
+                records.append({'车牌': k[0], '销售': k[1], '部门': g['部门'],
+                               '金额': bv, '类型': f'补发{bm}月(500)'})
+
+    records.sort(key=lambda x: (x['车牌'], x['类型']))
+    return records
+
+
+# ========== Excel输出 ==========
+
+def set_widths(ws, widths):
+    for i, w in enumerate(widths, 1):
+        ws.column_dimensions[openpyxl.utils.get_column_letter(i)].width = w
+
+
+def write_xlsx(recs_6, G6, pay_records, filepath):
+    wb = openpyxl.Workbook()
+
+    # Sheet 1: 规则
+    ws = wb.active
+    ws.title = '考核规则'
+    ws.append(['6月 18T 绩效考核 — 方案3'])
+    ws.append(['考核目标', '羚牛100辆18T'])
+    ws.append(['月度目标里程', '6000km'])
+    ws.append(['当月达标', '800元 × (考核天数/30天)'])
+    ws.append(['当月未达标', '300元 × (考核天数/30天)'])
+    ws.append(['结转奖励', '800元/次 (消耗1个可结转月)'])
+    ws.append(['补发历史月', '500元 × (历史月天数/历史总天数)'])
+    ws.append(['累计补发当月', '500元 × (当月天数/30天)'])
+    ws.append(['规则', '当月达标800/未达标300 → 结转800 → 补发500 → 累计补发500'])
+    ws.append(['亏损筛选', '6月无盈亏表，不管亏损全发'])
+    set_widths(ws, [30, 60])
+
+    # Sheet 2: 里程明细
+    ws2 = wb.create_sheet('6月里程明细')
+    ws2.append(['车牌号', '销售经理', '部门', '客户', '考核天数', '应考核里程', '实际里程', '是否达标', '月度目标'])
+    for r in recs_6:
+        ws2.append([r['车牌号'], r['销售经理'], r.get('部门名称', ''),
+                    r.get('客户名称', ''), r['考核天数'], r['应考核里程(km)'],
+                    r['实际行驶里程(km)'], r['是否达标'], TARGET_KM])
+    set_widths(ws2, [14, 10, 30, 35, 10, 12, 12, 10, 10])
+
+    # Sheet 3: 计算过程
+    ws3 = wb.create_sheet('6月计算过程')
+    ws3.append(['车牌号', '销售经理', '部门', '6月应考核', '6月实际', '6月达标',
+                '累计应考核', '累计实际', '累计达标', '可结转', '结转发放(800)',
+                '当月奖金', '累计补发6月(500)', '补发1月', '补发2月', '补发3月', '补发4月', '补发5月',
+                '6月总发放'])
+    for k, g in sorted(G6.items()):
+        if g.get('虚拟'):
+            continue
+        cum_q_str = '是' if g.get('cum_q') else '否'
+        dq_str = '是' if g['有达标'] else '否'
+        total = (g.get('结转', 0) + g.get('当月奖金', 0) + g.get('累计补发当月', 0) +
+                 sum(g.get(f'补发{m}月', 0) for m in range(1, 6)))
+        ws3.append([k[0], k[1], g['部门'], g['应考核'], g['实际'], dq_str,
+                    g.get('cum_t', ''), g.get('cum_a', ''), cum_q_str,
+                    g.get('可结转', 0), g.get('结转', 0),
+                    g.get('当月奖金', 0), g.get('累计补发当月', 0),
+                    g.get('补发1月', 0), g.get('补发2月', 0), g.get('补发3月', 0),
+                    g.get('补发4月', 0), g.get('补发5月', 0), total])
+    set_widths(ws3, [14, 10, 30, 12, 12, 10, 12, 12, 10, 10, 14, 12, 16, 10, 10, 10, 10, 10, 14])
+
+    # Sheet 4: 发放明细
+    ws4 = wb.create_sheet('6月发放明细')
+    ws4.append(['车牌号', '销售经理', '部门', '金额', '发放类型'])
+    for r in pay_records:
+        ws4.append([r['车牌'], r['销售'], r['部门'], round(r['金额'], 2), r['类型']])
+    total = sum(r['金额'] for r in pay_records)
+    ws4.append([])
+    ws4.append(['', '', '合计', round(total, 2), ''])
+    set_widths(ws4, [14, 10, 30, 12, 22])
+
+    # Sheet 5: 汇总
+    ws5 = wb.create_sheet('汇总')
+    ws5.append(['6月18T 方案3 汇总 (800/300, 结转800, 补发500)'])
+    ws5.append(['发放类型', '辆数', '金额合计'])
+    cat_sum, cat_cnt = defaultdict(float), defaultdict(int)
+    for r in pay_records:
+        cat_sum[r['类型']] += r['金额']
+        cat_cnt[r['类型']] += 1
+    all_cats = ['当月达标(800)', '当月未达标(300)', '结转(800)', '累计补发6月(500)',
+                '补发1月(500)', '补发2月(500)', '补发3月(500)', '补发4月(500)', '补发5月(500)']
+    for cat in all_cats:
+        if cat in cat_sum:
+            ws5.append([cat, cat_cnt[cat], round(cat_sum[cat], 2)])
+    ws5.append(['合计', sum(cat_cnt.values()), round(sum(cat_sum.values()), 2)])
+    set_widths(ws5, [22, 10, 15])
+
+    wb.save(filepath)
+    print(f'  ✅ {filepath}')
+
+
+# ========== 主流程 ==========
+
+def main():
+    os.makedirs(OUTPUT_DIR, exist_ok=True)
+
+    print('计算6月18T方案3...')
+    recs_6, G6, pay_records = calc()
+
+    # 统计
+    total_groups = sum(1 for g in G6.values() if not g.get('虚拟'))
+    good_groups = sum(1 for g in G6.values() if g['有达标'] and not g.get('虚拟'))
+    bad_groups = sum(1 for g in G6.values() if not g['有达标'] and not g.get('虚拟'))
+    carry_groups = sum(1 for g in G6.values() if g.get('结转', 0) > 0)
+
+    print(f'  6月18T组数: {total_groups}, 达标: {good_groups}, 未达标: {bad_groups}, 结转: {carry_groups}')
+
+    # 分类汇总
+    def cats(dl, *keys):
+        cnt = sum(sum(1 for r in dl if r['类型']==k) for k in keys)
+        amt = sum(sum(r['金额'] for r in dl if r['类型']==k) for k in keys)
+        return cnt, amt
+
+    def cell(amt, cnt):
+        if cnt == 0: return '—'
+        return f'{amt:,.2f} ({cnt}辆)'
+
+    rows = [
+        ('当月达标(800)',     cats(pay_records, '当月达标(800)')),
+        ('当月未达标(300)',   cats(pay_records, '当月未达标(300)')),
+        ('结转(800)',        cats(pay_records, '结转(800)')),
+        ('累计补发6月(500)', cats(pay_records, '累计补发6月(500)')),
+        ('补发1月(500)',     cats(pay_records, '补发1月(500)')),
+        ('补发2月(500)',     cats(pay_records, '补发2月(500)')),
+        ('补发3月(500)',     cats(pay_records, '补发3月(500)')),
+        ('补发4月(500)',     cats(pay_records, '补发4月(500)')),
+        ('补发5月(500)',     cats(pay_records, '补发5月(500)')),
+    ]
+    grand_total = sum(r['金额'] for r in pay_records)
+
+    def vw(s):
+        return sum(2 if unicodedata.east_asian_width(c) in ('F','W') else 1 for c in str(s))
+    def padr(s, w):
+        return ' ' * (w - vw(s)) + str(s)
+    def padl(s, w):
+        return str(s) + ' ' * (w - vw(s))
+
+    W0, W1 = 17, 26
+    print()
+    print('  6月 18T 绩效考核 — 方案3 (800/300, 结转800, 补发500)')
+    print(f"  ┌{'─'*W0}┬{'─'*W1}┐")
+    for label, (cnt, amt) in rows:
+        if amt > 0 or ('累计补发' in label) or ('补发' in label):
+            c = cell(amt, cnt)
+        else:
+            c = cell(amt, cnt)
+        if amt > 0 or cnt > 0:
+            print(f"  │{padl(label,W0)}│{padr(c,W1)}│")
+    # separator
+    print(f"  ├{'─'*W0}┼{'─'*W1}┤")
+    print(f"  │{padl('合计',W0)}│{padr(f'{grand_total:,.2f}',W1)}│")
+    print(f"  └{'─'*W0}┴{'─'*W1}┘")
+
+    # 写Excel
+    write_xlsx(recs_6, G6, pay_records, os.path.join(OUTPUT_DIR, '6月18T_方案3.xlsx'))
+    print(f"\n文件输出到: {OUTPUT_DIR}/")
+
+
+if __name__ == '__main__':
+    main()
diff --git a/main.py b/main.py
index f555cd7..a9417ab 100644
--- a/main.py
+++ b/main.py
@@ -1,40 +1,69 @@
 #!/usr/bin/env python3
-"""按月生成独立Excel文件 - 里程考核绩效（含亏损筛选）"""
+"""按月生成独立Excel文件 - 里程考核绩效，支持1-7月"""
 import os, openpyxl
 from calc_engine import *
 from excel_writer import *
 
 os.chdir('/Users/kkfluous/Downloads')
+JULY_SOURCE = os.environ.get('JULY_SOURCE', os.path.join('7月核算结果', '租赁任务考核_2026年7月.xlsx'))
+JULY_OUTPUT_DIR = os.environ.get('JULY_OUTPUT_DIR', '7月核算结果')
 
-# 读取
+# 读取1-7月；7月优先读取核算结果目录内已完成GPS干预的源表
 print("读取源数据...")
-D = {1: read_file('租赁任务考核_2026年1月.xlsx', 1),
-     2: read_file('租赁任务考核_2026年2月.xlsx', 2),
-     3: read_file('租赁任务考核_2026年3月.xlsx', 3)}
-for m in [1,2,3]: print(f"  {m}月: {len(D[m])}条")
+months = []
+for m in range(1, 8):
+    fp = JULY_SOURCE if m == 7 else f'租赁任务考核_2026年{m}月.xlsx'
+    if os.path.exists(fp):
+        months.append(m)
+D = {}
+for m in months:
+    fp = JULY_SOURCE if m == 7 else f'租赁任务考核_2026年{m}月.xlsx'
+    D[m] = read_file(fp, m)
+    print(f"  {m}月: {len(D[m])}条")
 
-# 分组+计算
-G = {m: grp(D[m], m) for m in [1,2,3]}
+# 分组
+G = {m: grp(D[m], m) for m in months}
+
+# 1月初始化
 calc_jan_carryover(G[1])
-feb_data = calc_feb(G[1], G[2])
-mar_data = calc_mar(G[1], G[2], G[3], feb_data)
+all_data = {}
+all_data[1] = {'当月': [{'车牌':k[0],'销售':g['销售'],'部门':g['部门'],'额':g['奖金']}
+                        for k,g in G[1].items() if g['奖金'] > 0]}
+
+# 逐月计算: 2→7
+for m in range(2, 8):
+    if m not in G:
+        continue
+    data_m, G[m] = calc_month(m, G, all_data, G[m-1], G[m])
+
+# 钱婧已离职：从2026年7月起所有结转、当月、补发及累计补发均不再发放。
+all_data[7] = {
+    cat: [item for item in items if item.get('销售') != '钱婧']
+    for cat, items in all_data[7].items()
+}
 
 # 发放追踪
-vehicle_payments, vehicle_info = collect_vehicle_payments(G, feb_data, mar_data)
+vehicle_payments, vehicle_info = collect_vehicle_payments(G, all_data)
 master_vehicles = read_master_vehicles()
 
 # 亏损表
 print("\n读取亏损表...")
 loss_data = {}
-for m in [1,2,3]:
+for m in months:
     ld = read_loss_data(m)
     if ld:
         loss_clients = sum(1 for v in ld.values() if v == '是')
         print(f"  {m}月: {len(ld)}个客户, 其中亏损{loss_clients}个")
         loss_data[m] = ld
     else:
-        print(f"  {m}月: 无亏损表")
-        loss_data[m] = None
+        # 无亏损表的月份不管亏损全发 → 构造全'否'的代理
+        from collections import defaultdict as dd
+        ld_tmp = dd(lambda: '否')
+        for r in D[m]:
+            c = r.get('客户名称')
+            if c: ld_tmp[c] = '否'
+        print(f"  {m}月: 无亏损表，按全部不亏损处理（全发）")
+        loss_data[m] = ld_tmp
 
 # 车牌→客户名称映射
 plate_client = get_vehicle_client_map(D)
@@ -46,27 +75,23 @@ for m in G:
         all_persons[g['销售']] = g['部门']
 
 # 打印汇总
-jan_total = sum(r['奖金'] for r in D[1] if r['是否达标']=='达标')
-feb_total = sum(sum(d['额'] for d in v) for v in feb_data.values())
-mar_total = sum(sum(d['额'] for d in v) for v in mar_data.values())
-print(f"\n考核应发: 1月{jan_total:.2f}, 2月{feb_total:.2f}, 3月{mar_total:.2f}")
-print(f"Q1考核应发合计: {jan_total + feb_total + mar_total:.2f}")
+print("\n考核应发汇总:")
+grand = 0
+for m in months:
+    total_m = sum(sum(d['额'] for d in v) for v in all_data[m].values())
+    grand += total_m
+    print(f"  {m}月: {total_m:.2f}")
+print(f"  合计: {grand:.2f}")
 
 # ============================================================
-# 按月生成
+# 按月生成（只生成7月）
 # ============================================================
-for settle_month in [1, 2, 3]:
+for settle_month in [7]:
     print(f"\n生成 {settle_month}月核算文件...")
     wb = openpyxl.Workbook()
 
     # 当月数据
-    if settle_month == 1:
-        month_data = {'当月': [{'车牌':r['车牌号'],'销售':r['销售经理'],'部门':r['部门名称'],'额':r['奖金']}
-                               for r in D[1] if r['是否达标']=='达标']}
-    elif settle_month == 2:
-        month_data = feb_data
-    else:
-        month_data = mar_data
+    month_data = all_data[settle_month]
 
     # Sheet 1: 考核奖励规则
     write_rules_sheet(wb)
@@ -75,31 +100,50 @@ for settle_month in [1, 2, 3]:
     write_detail_sheet(wb, D[settle_month], settle_month)
 
     # Sheet 3: 计算过程
-    if settle_month == 1:
-        write_calc_process_jan(wb, G[1])
-    elif settle_month == 2:
-        write_calc_process_feb(wb, G[1], G[2])
-    else:
-        write_calc_process_mar(wb, G[1], G[2], G[3], feb_data)
+    write_calc_process_generic(wb, settle_month, G, all_data)
 
     # Sheet 5: 车辆考核追踪
     write_vehicle_tracking_sheet(wb, settle_month, G, master_vehicles, vehicle_payments, vehicle_info, loss_data[settle_month], plate_client)
 
     # Sheet 6: 奖金发放记录（叠加亏损筛选，补发查对应月盈亏）
-    payment_records = build_payment_records(settle_month, month_data, loss_data, plate_client)
+    # 不做任何亏损拦截，全发
+    class _AllPass:
+        def get(self, _k, _d=None): return '否'
+        def __bool__(self): return True
+    payment_records = build_payment_records(settle_month, month_data, {m: _AllPass() for m in months}, plate_client)
     write_payment_record_sheet(wb, settle_month, payment_records)
 
-    # Sheet 7: 月汇总（从发放记录生成）
-    write_summary_from_records(wb, settle_month, payment_records)
+    # Sheet 7: 月汇总。离职人员仅当月奖金转给现销售，其余结转/补发不发；
+    # 发放记录和个人明细仍保留原归属，方便审计。
+    resignation_transfer = {'赵连飞': '刘念念', '伍仲文': '钟祥', '岑彦': '钟祥'}
+    summary_records = []
+    for rec in payment_records:
+        item = dict(rec)
+        sales = item.get('业务员') or item.get('销售人员') or item.get('销售')
+        ptype = item.get('发放类型') or item.get('类型')
+        if sales in resignation_transfer:
+            if ptype != '当月':
+                continue
+            new_sales = resignation_transfer[sales]
+            if '业务员' in item: item['业务员'] = new_sales
+            if '销售人员' in item: item['销售人员'] = new_sales
+            if '销售' in item: item['销售'] = new_sales
+        summary_records.append(item)
+    write_summary_from_records(wb, settle_month, summary_records)
 
-    # Sheet 6-17: 业务员
+    # 业务员sheets
     for person in sorted(all_persons.keys()):
         write_salesperson_sheet(wb, person, all_persons[person], settle_month, D, G, month_data, vehicle_payments)
 
-    # 删除默认空sheet
     if 'Sheet' in wb.sheetnames:
         del wb['Sheet']
 
-    fname = f'里程任务考核_{settle_month}月核算.xlsx'
+    for old in ['二部-赵连飞', '六部-伍仲文', '六部-岑彦', '三部-钱婧']:
+        if old in wb.sheetnames:
+            wb[old].title = f'【已离职】{old}'
+            wb.move_sheet(wb[f'【已离职】{old}'], offset=len(wb.sheetnames))
+
+    os.makedirs(JULY_OUTPUT_DIR, exist_ok=True)
+    fname = os.path.join(JULY_OUTPUT_DIR, f'里程任务考核_{settle_month}月核算.xlsx')
     wb.save(fname)
     print(f"  ✅ {fname} ({len(wb.sheetnames)} sheets)")
diff --git a/may_18t_recalc.py b/may_18t_recalc.py
new file mode 100644
index 0000000..3c02ab3
--- /dev/null
+++ b/may_18t_recalc.py
@@ -0,0 +1,741 @@
+#!/usr/bin/env python3
+"""
+5月 18T 绩效三种测算方案
+方案1: 保留原逻辑 (1000元基准, 含结转/补发/累计)
+方案2: 新测算简化版 (800/300, 仅当月, 无结转/补发/累计)
+方案3: 新测算完整版 (800/300/500, 含结转/补发/累计)
+"""
+
+import os, openpyxl
+from collections import defaultdict
+from copy import deepcopy
+
+os.chdir('/Users/kkfluous/Downloads')
+
+DAYS = {1: 31, 2: 28, 3: 31, 4: 30, 5: 31}
+TARGET_KM = 6000
+OUTPUT_DIR = '/Users/kkfluous/Downloads/5月18T测算'
+
+# ============================================================
+# 数据读取
+# ============================================================
+
+def read_18t(filepath, month):
+    """只读羚牛100辆18T记录"""
+    wb = openpyxl.load_workbook(filepath, data_only=True)
+    ws = wb['业务考核视图']
+    h = [c.value for c in next(ws.iter_rows(min_row=1, max_row=1))]
+    recs = []
+    for row in ws.iter_rows(min_row=2, values_only=True):
+        r = dict(zip(h, row))
+        if r.get('考核目标') != '羚牛100辆18T':
+            continue
+        r['考核天数'] = float(r.get('考核天数') or 0)
+        r['应考核里程(km)'] = float(r.get('应考核里程(km)') or 0)
+        r['实际行驶里程(km)'] = float(r.get('实际行驶里程(km)') or 0)
+        r['月份'] = month
+        t, a = r['应考核里程(km)'], r['实际行驶里程(km)']
+        q = a >= t and t > 0
+        r['是否达标'] = '达标' if q else '未达标'
+        r['多跑'] = max(0, a - t) if q else 0
+        r['可结转'] = int(r['多跑'] // TARGET_KM) if q else 0
+        recs.append(r)
+    wb.close()
+    return recs
+
+
+def grp(recs):
+    """按(车牌号, 销售经理)分组"""
+    gs = defaultdict(lambda: {
+        'recs': [], '应考核': 0, '实际': 0, '奖金': 0, '天数': 0,
+        '有达标': False, '目标km': TARGET_KM,
+        '部门': '', '销售': '', '车牌': ''
+    })
+    for r in recs:
+        k = (r['车牌号'], r['销售经理'])
+        g = gs[k]
+        g['recs'].append(r)
+        g['应考核'] += r['应考核里程(km)']
+        g['实际'] += r['实际行驶里程(km)']
+        g['奖金'] += r.get('奖金', 0)
+        g['天数'] += r['考核天数']
+        if r['是否达标'] == '达标':
+            g['有达标'] = True
+        g['部门'] = r['部门名称']
+        g['销售'] = r['销售经理']
+        g['车牌'] = r['车牌号']
+    return dict(gs)
+
+
+# ============================================================
+# 方案1: 原逻辑 (base=1000)
+# ============================================================
+
+def calc_s1():
+    """方案1: 保留原逻辑，base=1000。只算18T"""
+    BASE = 1000
+    D = {}
+    for m in range(1, 6):
+        fp = f'租赁任务考核_2026年{m}月.xlsx'
+        if os.path.exists(fp):
+            D[m] = read_18t(fp, m)
+
+    # 给每条记录加上奖励额并计算奖金
+    for m in D:
+        for r in D[m]:
+            r['奖励额'] = BASE
+            t, a = r['应考核里程(km)'], r['实际行驶里程(km)']
+            q = a >= t and t > 0
+            r['是否达标'] = '达标' if q else '未达标'
+            r['奖金'] = BASE * (r['考核天数'] / DAYS[m]) if q else 0
+            r['多跑'] = max(0, a - t) if q else 0
+            r['可结转'] = int(r['多跑'] // TARGET_KM) if q else 0
+
+    G = {m: grp(D[m]) for m in D}
+
+    # 1月: 计算可结转 + 初始化累计
+    for k, g in G[1].items():
+        cum_q = g['实际'] >= g['应考核'] and g['应考核'] > 0
+        cum_over = max(0, g['实际'] - g['应考核']) if cum_q else 0
+        g['可结转'] = int(cum_over // TARGET_KM) if TARGET_KM > 0 else 0
+        g['累计已发结转'] = 0
+        g['1月已发'] = g['奖金'] > 0
+        g['奖励额'] = BASE
+        g['cum_t'] = g['应考核']
+        g['cum_a'] = g['实际']
+        g['cum_q'] = cum_q
+
+    # 用于追踪每个月的发放
+    all_data = {1: {'结转': [], '当月': []}}
+
+    # 用于追踪每个月的发放
+    all_data = {1: {'结转': [], '当月': []}}
+
+    for k, g in G[1].items():
+        if g['奖金'] > 0:
+            all_data[1]['当月'].append({'车牌': k[0], '销售': g['销售'], '部门': g['部门'], '额': g['奖金']})
+
+    # 逐月计算: 2→5
+    for month in range(2, 6):
+        if month not in G:
+            continue
+        data_m = {'结转': [], '当月': []}
+        prev_m = month - 1
+        days_m = DAYS[month]
+
+        for k, gm in G[month].items():
+            gp = G[prev_m].get(k)
+            bf = BASE
+
+            # 累计
+            cum_t = (gp['cum_t'] if gp else 0) + gm['应考核']
+            cum_a = (gp['cum_a'] if gp else 0) + gm['实际']
+            cum_q = cum_a >= cum_t and cum_t > 0
+
+            # 结转
+            carry = 0
+            p_carry = gp['可结转'] if gp else 0
+            if gp and p_carry >= 1:
+                carry = bf
+                data_m['结转'].append({'车牌': k[0], '销售': gp['销售'], '部门': gm['部门'], '额': carry})
+
+            # 当月达标
+            bonus = 0
+            if gm['有达标'] and carry == 0:
+                bonus = gm['奖金']
+                data_m['当月'].append({'车牌': k[0], '销售': gm['销售'], '部门': gm['部门'], '额': bonus})
+
+            # 补发历史月
+            for back_m in range(1, month):
+                label = f'补发{back_m}月'
+                if label not in data_m:
+                    data_m[label] = []
+
+            for back_m in range(1, month):
+                label = f'补发{back_m}月'
+                g_bm = G[back_m].get(k)
+                if not g_bm:
+                    continue
+                # 检查该月是否已发
+                bm_paid = g_bm.get(f'{back_m}月已发', g_bm['奖金'] > 0)
+                if bm_paid:
+                    continue
+                # 检查后续月是否已补发
+                bm_paid_later = False
+                for later_m in range(back_m + 1, month):
+                    gl = G[later_m].get(k)
+                    if gl and gl.get(f'补发{back_m}月', 0) > 0:
+                        bm_paid_later = True
+                        break
+                if bm_paid_later:
+                    continue
+                if cum_q:
+                    bj = g_bm['奖励额'] * (g_bm['天数'] / DAYS[back_m])
+                    data_m[label].append({'车牌': k[0], '销售': g_bm['销售'], '部门': gm['部门'], '额': bj})
+                    gm[f'补发{back_m}月'] = bj
+
+            # 累计补发当月
+            label_cbp = f'累计补发{month}月'
+            if label_cbp not in data_m:
+                data_m[label_cbp] = []
+            cbp = 0
+            if not gm['有达标'] and carry == 0 and cum_q:
+                cbp = bf * (gm['天数'] / days_m)
+                data_m[label_cbp].append({'车牌': k[0], '销售': gm['销售'], '部门': gm['部门'], '额': cbp})
+
+            gm['cum_t'] = cum_t
+            gm['cum_a'] = cum_a
+            gm['cum_q'] = cum_q
+            gm['结转'] = carry
+            gm['当月奖金'] = bonus
+            gm['累计补发当月'] = cbp
+            gm['奖励额'] = BASE
+            gm[f'{month}月已发'] = carry > 0 or bonus > 0 or cbp > 0
+
+            # 更新可结转
+            cum_over = max(0, cum_a - cum_t) if cum_q else 0
+            earned = int(cum_over // TARGET_KM) if TARGET_KM > 0 else 0
+            prev_used = gp['累计已发结转'] if gp else 0
+            cur_used = prev_used + (1 if carry > 0 else 0)
+            gm['累计已发结转'] = cur_used
+            gm['可结转'] = max(0, earned - cur_used)
+
+        # 补充: 上个月有结转但本月无考核记录的车
+        for k, gp in G[prev_m].items():
+            if k not in G[month] and gp.get('可结转', 0) >= 1:
+                carry = BASE
+                data_m['结转'].append({'车牌': k[0], '销售': gp['销售'], '部门': gp['部门'], '额': carry})
+                G[month][k] = {
+                    'recs': [], '应考核': TARGET_KM, '实际': 0, '奖金': 0, '天数': days_m,
+                    '有达标': False, '目标km': TARGET_KM, '部门': gp['部门'], '销售': gp['销售'],
+                    '车牌': k[0], '奖励额': BASE,
+                    'cum_t': gp.get('cum_t', 0) + TARGET_KM, 'cum_a': gp.get('cum_a', 0),
+                    'cum_q': False,
+                    '结转': carry, '当月奖金': 0,
+                    f'{month}月已发': True,
+                    '可结转': max(0, gp['可结转'] - 1),
+                    '累计已发结转': gp.get('累计已发结转', 0) + 1,
+                    '虚拟': True,
+                }
+                for bm in range(1, month):
+                    G[month][k][f'补发{bm}月'] = 0
+
+        all_data[month] = data_m
+
+    # 收集5月发放明细
+    pay_records = collect_payments_s1(G, all_data)
+
+    return D[5], G[5], all_data.get(5, {}), pay_records
+
+
+def collect_payments_s1(G, all_data):
+    """收集5月结算的所有发放记录"""
+    records = []
+    m5 = all_data.get(5, {})
+    for cat, items in m5.items():
+        for d in items:
+            records.append({
+                '车牌': d['车牌'], '销售': d['销售'], '部门': d.get('部门', ''),
+                '金额': d['额'], '类型': cat, '结算月': 5
+            })
+    records.sort(key=lambda x: (x['车牌'], x['类型']))
+    return records
+
+
+# ============================================================
+# 方案2: 简化版 (800/300, 仅当月)
+# ============================================================
+
+def calc_s2():
+    """方案2: 仅看5月, 达标800, 未达标300, 无结转/补发/累计。按组统计(同方案1)"""
+    BASE_GOOD = 800
+    BASE_BAD = 300
+
+    fp = '租赁任务考核_2026年5月.xlsx'
+    recs = read_18t(fp, 5)
+
+    # 给每条记录算好奖金
+    for r in recs:
+        if r['是否达标'] == '达标':
+            r['奖金'] = BASE_GOOD * (r['考核天数'] / DAYS[5])
+        else:
+            r['奖金'] = BASE_BAD * (r['考核天数'] / DAYS[5])
+
+    # 按(车牌+销售)分组, 每组一笔 (同方案1)
+    G5 = grp(recs)
+    pay_records = []
+    for k, g in sorted(G5.items()):
+        total = g['奖金']  # 组内达标+未达标奖金合计
+        ptype = '当月达标(800)' if g['有达标'] else '当月未达标(300)'
+        pay_records.append({
+            '车牌': k[0], '销售': k[1], '部门': g['部门'],
+            '金额': total, '类型': ptype
+        })
+
+    pay_records.sort(key=lambda x: (x['车牌'], x['销售']))
+    return recs, pay_records
+
+
+# ============================================================
+# 方案3: 完整版 (800/300, 结转800, 补发/累计500)
+# ============================================================
+
+def calc_s3():
+    """方案3: 当月达标800, 当月未达标300, 结转800, 补发/累计补发500"""
+    BASE_GOOD = 800
+    BASE_BAD = 300
+    BASE_CARRY = 800
+    BASE_OTHER = 500
+
+    D = {}
+    for m in range(1, 6):
+        fp = f'租赁任务考核_2026年{m}月.xlsx'
+        if os.path.exists(fp):
+            D[m] = read_18t(fp, m)
+
+    # 给每条记录加上奖励额和当月奖金(按新标准)
+    for m in D:
+        for r in D[m]:
+            r['奖励额'] = BASE_GOOD  # 当月达标基准
+            r['奖励额_未达标'] = BASE_BAD  # 当月未达标基准
+            r['奖励额_其他'] = BASE_OTHER  # 结转/补发/累计基准
+            t, a = r['应考核里程(km)'], r['实际行驶里程(km)']
+            q = a >= t and t > 0
+            r['是否达标'] = '达标' if q else '未达标'
+            if q:
+                r['奖金'] = BASE_GOOD * (r['考核天数'] / DAYS[m])
+            else:
+                r['奖金'] = BASE_BAD * (r['考核天数'] / DAYS[m])
+            r['多跑'] = max(0, a - t) if q else 0
+            r['可结转'] = int(r['多跑'] // TARGET_KM) if q else 0
+
+    G = {m: grp(D[m]) for m in D}
+
+    # 1月: 计算可结转 + 初始化累计
+    for k, g in G[1].items():
+        g['奖励额'] = BASE_GOOD
+        g['奖励额_其他'] = BASE_OTHER
+        cum_q = g['实际'] >= g['应考核'] and g['应考核'] > 0
+        cum_over = max(0, g['实际'] - g['应考核']) if cum_q else 0
+        g['可结转'] = int(cum_over // TARGET_KM) if TARGET_KM > 0 else 0
+        g['累计已发结转'] = 0
+        # 方案3: 1月已发 = 有达标 (仅300未达标不算)
+        g['1月已发'] = g['有达标']
+        g['cum_t'] = g['应考核']
+        g['cum_a'] = g['实际']
+        g['cum_q'] = cum_q
+
+    all_data = {1: {'结转': [], '当月达标(800)': [], '当月未达标(300)': []}}
+    for k, g in G[1].items():
+        # 按组统计: 有达标→当月达标(合并金额), 全未达标→当月未达标
+        if g['有达标']:
+            all_data[1]['当月达标(800)'].append({'车牌': k[0], '销售': g['销售'], '部门': g['部门'], '额': g['奖金']})
+        elif g['奖金'] > 0:
+            all_data[1]['当月未达标(300)'].append({'车牌': k[0], '销售': g['销售'], '部门': g['部门'], '额': g['奖金']})
+
+    # 逐月计算: 2→5
+    for month in range(2, 6):
+        if month not in G:
+            continue
+        data_m = {'结转': [], '当月达标(800)': [], '当月未达标(300)': []}
+        prev_m = month - 1
+        days_m = DAYS[month]
+
+        for k, gm in G[month].items():
+            gp = G[prev_m].get(k)
+
+            # 累计
+            cum_t = (gp['cum_t'] if gp else 0) + gm['应考核']
+            cum_a = (gp['cum_a'] if gp else 0) + gm['实际']
+            cum_q = cum_a >= cum_t and cum_t > 0
+
+            # 结转: 800 (整月)
+            carry = 0
+            p_carry = gp['可结转'] if gp else 0
+            if gp and p_carry >= 1:
+                carry = BASE_CARRY
+                data_m['结转'].append({'车牌': k[0], '销售': gp['销售'], '部门': gm['部门'], '额': carry})
+
+            # 当月奖金: 按组统计(同方案1), 有达标→当月达标, 全未达标→当月未达标
+            if carry == 0 and gm['奖金'] > 0:
+                ptype = '当月达标(800)' if gm['有达标'] else '当月未达标(300)'
+                data_m[ptype].append({'车牌': k[0], '销售': gm['销售'], '部门': gm['部门'], '额': gm['奖金']})
+
+            # 补发历史月: 500基准
+            for back_m in range(1, month):
+                label = f'补发{back_m}月'
+                if label not in data_m:
+                    data_m[label] = []
+                g_bm = G[back_m].get(k)
+                if not g_bm:
+                    continue
+                # 方案3: 历史月已发 = 有达标/已结转/已补发/已累计补发. 仅拿300未达标不算已发.
+                bm_paid = g_bm.get(f'{back_m}月已发', g_bm['有达标'])
+                if bm_paid:
+                    continue
+                bm_paid_later = False
+                for later_m in range(back_m + 1, month):
+                    gl = G[later_m].get(k)
+                    if gl and gl.get(f'补发{back_m}月', 0) > 0:
+                        bm_paid_later = True
+                        break
+                if bm_paid_later:
+                    continue
+                if cum_q:
+                    bj = BASE_OTHER * (g_bm['天数'] / DAYS[back_m])
+                    data_m[label].append({'车牌': k[0], '销售': g_bm['销售'], '部门': gm['部门'], '额': bj})
+                    gm[f'补发{back_m}月'] = bj
+
+            # 累计补发当月: 500基准
+            label_cbp = f'累计补发{month}月'
+            if label_cbp not in data_m:
+                data_m[label_cbp] = []
+            cbp = 0
+            if not gm['有达标'] and carry == 0 and cum_q:
+                cbp = BASE_OTHER * (gm['天数'] / days_m)
+                data_m[label_cbp].append({'车牌': k[0], '销售': gm['销售'], '部门': gm['部门'], '额': cbp})
+
+            gm['cum_t'] = cum_t
+            gm['cum_a'] = cum_a
+            gm['cum_q'] = cum_q
+            gm['结转'] = carry
+            gm['当月奖金'] = gm['奖金'] if carry == 0 else 0
+            gm['累计补发当月'] = cbp
+            gm['奖励额'] = BASE_GOOD
+            gm['奖励额_其他'] = BASE_OTHER
+            # 方案3: 已发 = 结转/当月有达标/累计补发 (仅300未达标不算已发，可后续补发)
+            gm[f'{month}月已发'] = carry > 0 or gm['有达标'] or cbp > 0
+
+            # 更新可结转
+            cum_over = max(0, cum_a - cum_t) if cum_q else 0
+            earned = int(cum_over // TARGET_KM) if TARGET_KM > 0 else 0
+            prev_used = gp['累计已发结转'] if gp else 0
+            cur_used = prev_used + (1 if carry > 0 else 0)
+            gm['累计已发结转'] = cur_used
+            gm['可结转'] = max(0, earned - cur_used)
+
+        # 补充: 上个月有结转但本月无考核记录的车
+        for k, gp in G[prev_m].items():
+            if k not in G[month] and gp.get('可结转', 0) >= 1:
+                carry = BASE_CARRY
+                data_m['结转'].append({'车牌': k[0], '销售': gp['销售'], '部门': gp['部门'], '额': carry})
+                G[month][k] = {
+                    'recs': [], '应考核': TARGET_KM, '实际': 0, '奖金': 0, '天数': days_m,
+                    '有达标': False, '目标km': TARGET_KM, '部门': gp['部门'], '销售': gp['销售'],
+                    '车牌': k[0], '奖励额': BASE_GOOD, '奖励额_其他': BASE_OTHER,
+                    'cum_t': gp.get('cum_t', 0) + TARGET_KM, 'cum_a': gp.get('cum_a', 0),
+                    'cum_q': False,
+                    '结转': carry, '当月奖金': 0,
+                    f'{month}月已发': True,
+                    '可结转': max(0, gp['可结转'] - 1),
+                    '累计已发结转': gp.get('累计已发结转', 0) + 1,
+                    '虚拟': True,
+                }
+                for bm in range(1, month):
+                    G[month][k][f'补发{bm}月'] = 0
+
+        all_data[month] = data_m
+
+    pay_records = collect_payments_s3(G, all_data)
+
+    return D[5], G[5], all_data.get(5, {}), pay_records
+
+
+def collect_payments_s3(G, all_data):
+    """收集方案3的5月发放记录"""
+    records = []
+    m5 = all_data.get(5, {})
+    for cat, items in m5.items():
+        for d in items:
+            records.append({
+                '车牌': d['车牌'], '销售': d['销售'], '部门': d.get('部门', ''),
+                '金额': d['额'], '类型': cat, '结算月': 5
+            })
+    records.sort(key=lambda x: (x['车牌'], x['类型']))
+    return records
+
+
+# ============================================================
+# Excel 输出
+# ============================================================
+
+def write_xlsx_s1(recs_5, G5, data5, pay_records, filepath):
+    """方案1输出"""
+    wb = openpyxl.Workbook()
+
+    # Sheet 1: 规则
+    ws = wb.active
+    ws.title = '考核规则'
+    ws.append(['方案1: 保留原逻辑'])
+    ws.append(['考核目标', '羚牛100辆18T'])
+    ws.append(['月度目标里程', '6000km'])
+    ws.append(['奖励金额(达标)', '1000元/月 (按考核天数/当月天数折算)'])
+    ws.append(['结转奖励', '1000元/次 (消耗1个月目标)'])
+    ws.append(['补发/累计补发', '1000元/月 (按考核天数/当月天数折算)'])
+    ws.append(['规则', '达标→1000, 结转→1000, 补发→1000×折算, 累计补发当月→1000×折算'])
+    set_col_widths(ws, [30, 60])
+
+    # Sheet 2: 5月里程明细
+    write_mileage_detail(wb, recs_5, '5月里程明细')
+
+    # Sheet 3: 计算过程
+    ws3 = wb.create_sheet('5月计算过程')
+    ws3.append(['车牌号', '销售经理', '部门', '5月应考核', '5月实际', '5月达标',
+                '累计应考核', '累计实际', '累计达标', '可结转', '结转发放',
+                '当月奖金', '累计补发5月', '补发1月', '补发2月', '补发3月', '补发4月',
+                '5月总发放'])
+    for k, g in sorted(G5.items()):
+        if g.get('虚拟'):
+            continue
+        cum_q_str = '是' if g.get('cum_q') else '否'
+        dq_str = '是' if g['有达标'] else '否'
+        total = (g.get('结转', 0) + g.get('当月奖金', 0) + g.get('累计补发当月', 0) +
+                 g.get('补发1月', 0) + g.get('补发2月', 0) + g.get('补发3月', 0) +
+                 g.get('补发4月', 0))
+        ws3.append([k[0], k[1], g['部门'], g['应考核'], g['实际'], dq_str,
+                    g.get('cum_t', ''), g.get('cum_a', ''), cum_q_str,
+                    g.get('可结转', 0), g.get('结转', 0),
+                    g.get('当月奖金', 0), g.get('累计补发当月', 0),
+                    g.get('补发1月', 0), g.get('补发2月', 0), g.get('补发3月', 0),
+                    g.get('补发4月', 0), total])
+    set_col_widths(ws3, [14, 10, 30, 12, 12, 10, 12, 12, 10, 10, 12, 12, 14, 12, 12, 12, 12, 14])
+
+    # Sheet 4: 发放明细
+    ws4 = wb.create_sheet('5月发放明细')
+    ws4.append(['车牌号', '销售经理', '部门', '金额', '发放类型'])
+    for r in pay_records:
+        ws4.append([r['车牌'], r['销售'], r['部门'], round(r['金额'], 2), r['类型']])
+    # 汇总行
+    total = sum(r['金额'] for r in pay_records)
+    ws4.append([])
+    ws4.append(['', '', '合计', round(total, 2), ''])
+    set_col_widths(ws4, [14, 10, 30, 12, 20])
+
+    # Sheet 5: 汇总
+    ws5 = wb.create_sheet('汇总')
+    ws5.append(['方案1: 原逻辑 (1000元基准)'])
+    ws5.append(['发放类型', '辆数', '金额合计'])
+    from collections import defaultdict
+    cat_sum = defaultdict(float)
+    cat_cnt = defaultdict(int)
+    for r in pay_records:
+        cat_sum[r['类型']] += r['金额']
+        cat_cnt[r['类型']] += 1
+    for cat in ['当月', '结转', '累计补发5月', '补发1月', '补发2月', '补发3月', '补发4月']:
+        if cat in cat_sum:
+            ws5.append([cat, cat_cnt[cat], round(cat_sum[cat], 2)])
+    ws5.append(['合计', sum(cat_cnt.values()), round(sum(cat_sum.values()), 2)])
+    set_col_widths(ws5, [20, 10, 15])
+
+    wb.save(filepath)
+    print(f'  ✅ {filepath}')
+
+
+def write_xlsx_s2(recs_5, pay_records, filepath):
+    """方案2输出"""
+    wb = openpyxl.Workbook()
+
+    # Sheet 1: 规则
+    ws = wb.active
+    ws.title = '考核规则'
+    ws.append(['方案2: 新测算简化版 (仅当月)'])
+    ws.append(['考核目标', '羚牛100辆18T'])
+    ws.append(['月度目标里程', '6000km'])
+    ws.append(['当月达标', '800元 × (考核天数/31天)'])
+    ws.append(['当月未达标', '300元 × (考核天数/31天)'])
+    ws.append(['结转/补发/累计', '不计算'])
+    ws.append(['规则', '只看5月当月，按(车牌+销售)分组统计'])
+    set_col_widths(ws, [30, 60])
+
+    # Sheet 2: 5月里程明细
+    write_mileage_detail(wb, recs_5, '5月里程明细')
+
+    # Sheet 3: 发放明细
+    ws3 = wb.create_sheet('5月发放明细')
+    ws3.append(['车牌号', '销售经理', '部门', '金额', '发放类型'])
+    for r in pay_records:
+        ws3.append([r['车牌'], r['销售'], r['部门'], round(r['金额'], 2), r['类型']])
+    total = sum(r['金额'] for r in pay_records)
+    ws3.append([])
+    ws3.append(['', '', '合计', round(total, 2), ''])
+    set_col_widths(ws3, [14, 10, 30, 12, 20])
+
+    # Sheet 4: 汇总
+    ws4 = wb.create_sheet('汇总')
+    ws4.append(['方案2: 简化版 (800/300, 仅当月)'])
+    ws4.append(['发放类型', '辆数', '金额合计'])
+    cat_sum = defaultdict(float)
+    cat_cnt = defaultdict(int)
+    for r in pay_records:
+        cat_sum[r['类型']] += r['金额']
+        cat_cnt[r['类型']] += 1
+    for cat in sorted(cat_sum.keys()):
+        ws4.append([cat, cat_cnt[cat], round(cat_sum[cat], 2)])
+    ws4.append(['合计', sum(cat_cnt.values()), round(sum(cat_sum.values()), 2)])
+    set_col_widths(ws4, [25, 10, 15])
+
+    wb.save(filepath)
+    print(f'  ✅ {filepath}')
+
+
+def write_xlsx_s3(recs_5, G5, data5, pay_records, filepath):
+    """方案3输出"""
+    wb = openpyxl.Workbook()
+
+    # Sheet 1: 规则
+    ws = wb.active
+    ws.title = '考核规则'
+    ws.append(['方案3: 新测算完整版 (800/300/500, 结转800 补发500)'])
+    ws.append(['考核目标', '羚牛100辆18T'])
+    ws.append(['月度目标里程', '6000km'])
+    ws.append(['当月达标', '800元 × (考核天数/31天)'])
+    ws.append(['当月未达标', '300元 × (考核天数/31天)'])
+    ws.append(['结转奖励', '800元/次 (消耗1个月目标)'])
+    ws.append(['补发历史月', '500元 × (历史月考核天数/历史月天数)'])
+    ws.append(['累计补发当月', '500元 × (当月考核天数/31天)'])
+    ws.append(['规则', '当月达标800/未达标300 → 结转800 → 补发500 → 累计补发500'])
+    set_col_widths(ws, [30, 60])
+
+    # Sheet 2: 5月里程明细
+    write_mileage_detail(wb, recs_5, '5月里程明细')
+
+    # Sheet 3: 计算过程
+    ws3 = wb.create_sheet('5月计算过程')
+    ws3.append(['车牌号', '销售经理', '部门', '5月应考核', '5月实际', '5月达标',
+                '累计应考核', '累计实际', '累计达标', '可结转', '结转发放(800)',
+                '当月奖金', '累计补发5月(500)', '补发1月', '补发2月', '补发3月', '补发4月',
+                '5月总发放'])
+    for k, g in sorted(G5.items()):
+        if g.get('虚拟'):
+            continue
+        cum_q_str = '是' if g.get('cum_q') else '否'
+        dq_str = '是' if g['有达标'] else '否'
+        total = (g.get('结转', 0) + g.get('当月奖金', 0) + g.get('累计补发当月', 0) +
+                 g.get('补发1月', 0) + g.get('补发2月', 0) + g.get('补发3月', 0) +
+                 g.get('补发4月', 0))
+        ws3.append([k[0], k[1], g['部门'], g['应考核'], g['实际'], dq_str,
+                    g.get('cum_t', ''), g.get('cum_a', ''), cum_q_str,
+                    g.get('可结转', 0), g.get('结转', 0),
+                    g.get('当月奖金', 0), g.get('累计补发当月', 0),
+                    g.get('补发1月', 0), g.get('补发2月', 0), g.get('补发3月', 0),
+                    g.get('补发4月', 0), total])
+    set_col_widths(ws3, [14, 10, 30, 12, 12, 10, 12, 12, 10, 10, 14, 12, 16, 12, 12, 12, 12, 14])
+
+    # Sheet 4: 发放明细
+    ws4 = wb.create_sheet('5月发放明细')
+    ws4.append(['车牌号', '销售经理', '部门', '金额', '发放类型'])
+    for r in pay_records:
+        ws4.append([r['车牌'], r['销售'], r['部门'], round(r['金额'], 2), r['类型']])
+    total = sum(r['金额'] for r in pay_records)
+    ws4.append([])
+    ws4.append(['', '', '合计', round(total, 2), ''])
+    set_col_widths(ws4, [14, 10, 30, 12, 20])
+
+    # Sheet 5: 汇总
+    ws5 = wb.create_sheet('汇总')
+    ws5.append(['方案3: 完整版 (800/300/500)'])
+    ws5.append(['发放类型', '辆数', '金额合计'])
+    cat_sum = defaultdict(float)
+    cat_cnt = defaultdict(int)
+    for r in pay_records:
+        cat_sum[r['类型']] += r['金额']
+        cat_cnt[r['类型']] += 1
+    for cat in ['当月达标(800)', '当月未达标(300)', '结转', '累计补发5月', '补发1月', '补发2月', '补发3月', '补发4月']:
+        if cat in cat_sum:
+            ws5.append([cat, cat_cnt[cat], round(cat_sum[cat], 2)])
+    ws5.append(['合计', sum(cat_cnt.values()), round(sum(cat_sum.values()), 2)])
+    set_col_widths(ws5, [20, 10, 15])
+
+    wb.save(filepath)
+    print(f'  ✅ {filepath}')
+
+
+def write_mileage_detail(wb, recs, sheet_name):
+    """里程明细sheet"""
+    ws = wb.create_sheet(sheet_name)
+    ws.append(['车牌号', '销售经理', '部门', '客户', '考核天数', '应考核里程',
+               '实际里程', '是否达标', '月度目标'])
+    for r in recs:
+        ws.append([r['车牌号'], r['销售经理'], r.get('部门名称', ''),
+                   r.get('客户名称', ''), r['考核天数'], r['应考核里程(km)'],
+                   r['实际行驶里程(km)'], r['是否达标'], TARGET_KM])
+    set_col_widths(ws, [14, 10, 30, 35, 10, 12, 12, 10, 10])
+
+
+def set_col_widths(ws, widths):
+    for i, w in enumerate(widths, 1):
+        ws.column_dimensions[openpyxl.utils.get_column_letter(i)].width = w
+
+
+# ============================================================
+# 主流程
+# ============================================================
+
+def main():
+    os.makedirs(OUTPUT_DIR, exist_ok=True)
+
+    # 所有计算
+    recs_5, G5_s1, data5_s1, pay_s1 = calc_s1()
+    recs_5_s2, pay_s2 = calc_s2()
+    recs_5_s3, G5_s3, data5_s3, pay_s3 = calc_s3()
+
+    # 汇总各方案
+    import unicodedata
+    def vw(s):
+        return sum(2 if unicodedata.east_asian_width(c) in ('F','W') else 1 for c in str(s))
+    def padr(s, w):
+        return ' ' * (w - vw(s)) + str(s)
+    def padl(s, w):
+        return str(s) + ' ' * (w - vw(s))
+
+    def cats(dl, *keys):
+        cnt = sum(sum(1 for r in dl if r['类型']==k) for k in keys)
+        amt = sum(sum(r['金额'] for r in dl if r['类型']==k) for k in keys)
+        return cnt, amt
+
+    def cell(amt, cnt):
+        if cnt == 0: return '—'
+        return f'{amt:,.2f} ({cnt}辆)'
+
+    rows = [
+        ('当月达标',     cats(pay_s1, '当月'),        cats(pay_s2, '当月达标(800)'),  cats(pay_s3, '当月达标(800)')),
+        ('当月未达标',   cats(pay_s1),                cats(pay_s2, '当月未达标(300)'), cats(pay_s3, '当月未达标(300)')),
+        ('结转',        cats(pay_s1, '结转'),         cats(pay_s2),                   cats(pay_s3, '结转')),
+        ('累计补发当月', cats(pay_s1, '累计补发5月'),  cats(pay_s2),                   cats(pay_s3, '累计补发5月')),
+        ('补发2月',     cats(pay_s1, '补发2月'),      cats(pay_s2),                   cats(pay_s3, '补发2月')),
+        ('补发3月',     cats(pay_s1, '补发3月'),      cats(pay_s2),                   cats(pay_s3, '补发3月')),
+        ('补发4月',     cats(pay_s1, '补发4月'),      cats(pay_s2),                   cats(pay_s3, '补发4月')),
+    ]
+    t1 = sum(r['金额'] for r in pay_s1)
+    t2 = sum(r['金额'] for r in pay_s2)
+    t3 = sum(r['金额'] for r in pay_s3)
+
+    H1, H2, H3 = '方案1: 原逻辑 (1000)', '方案2: 简化 (800/300)', '方案3: 完整 (结转800 补发500)'
+    W0, W1, W2, W3 = 15, 24, 24, 27
+
+    print()
+    print('  三方案对比')
+    print(f"  ┌{'─'*W0}┬{'─'*W1}┬{'─'*W2}┬{'─'*W3}┐")
+    print(f"  │{padl('',W0)}│{padl(H1,W1)}│{padl(H2,W2)}│{padl(H3,W3)}│")
+    for i, (label, vs1, vs2, vs3) in enumerate(rows):
+        if i == 2:
+            print(f"  ├{'─'*W0}┼{'─'*W1}┼{'─'*W2}┼{'─'*W3}┤")
+        c1 = cell(vs1[1], vs1[0]); c2 = cell(vs2[1], vs2[0]); c3 = cell(vs3[1], vs3[0])
+        print(f"  │{padl(label,W0)}│{padr(c1,W1)}│{padr(c2,W2)}│{padr(c3,W3)}│")
+    print(f"  ├{'─'*W0}┼{'─'*W1}┼{'─'*W2}┼{'─'*W3}┤")
+    print(f"  │{padl('合计',W0)}│{padr(f'{t1:,.2f}',W1)}│{padr(f'{t2:,.2f}',W2)}│{padr(f'{t3:,.2f}',W3)}│")
+    print(f"  └{'─'*W0}┴{'─'*W1}┴{'─'*W2}┴{'─'*W3}┘")
+
+    # 写Excel
+    print("\n生成Excel文件...")
+    write_xlsx_s1(recs_5, G5_s1, data5_s1, pay_s1,
+                  os.path.join(OUTPUT_DIR, '方案1_原逻辑_1000基准.xlsx'))
+    write_xlsx_s2(recs_5_s2, pay_s2,
+                  os.path.join(OUTPUT_DIR, '方案2_简化版_800_300仅当月.xlsx'))
+    write_xlsx_s3(recs_5_s3, G5_s3, data5_s3, pay_s3,
+                  os.path.join(OUTPUT_DIR, '方案3_完整版_800_300_500.xlsx'))
+    print(f"文件输出到: {OUTPUT_DIR}/")
+
+
+if __name__ == '__main__':
+    main()
diff --git a/openapi.yaml b/openapi.yaml
new file mode 100644
index 0000000..aa42405
--- /dev/null
+++ b/openapi.yaml
@@ -0,0 +1,553 @@
+openapi: 3.0.3
+info:
+  title: 车辆数据开放平台 API
+  version: 1.1.0
+  license:
+    name: Proprietary
+  description: |
+    向授权合作方开放车辆单日用氢量、单日里程和指定时刻总里程。
+    appKey 和逐车授权必须完整覆盖查询自然日。
+servers:
+  - url: /
+    description: 当前服务
+tags:
+  - name: 合作方数据接口
+    description: 使用 appKey 查询已授权车辆的日统计数据
+  - name: 开放平台管理
+    description: 仅车辆数据平台管理员可调用的应用和车辆授权管理接口
+paths:
+  /api/v1/vehicles/hydrogen-consumption/query:
+    post:
+      tags: [合作方数据接口]
+      summary: 查询车辆单日用氢量
+      operationId: queryDailyHydrogenConsumption
+      security:
+        - AppKeyAuth: []
+      requestBody:
+        required: true
+        content:
+          application/json:
+            schema:
+              $ref: '#/components/schemas/VehicleDailyQuery'
+            example:
+              plateNumbers: [粤A12345, 粤B67890]
+              date: '2026-07-01'
+      responses:
+        '200':
+          description: 查询成功；无数据车辆仍保留在结果中
+          content:
+            application/json:
+              schema:
+                $ref: '#/components/schemas/HydrogenQueryResponse'
+        '400':
+          $ref: '#/components/responses/BadRequest'
+        '401':
+          $ref: '#/components/responses/Unauthorized'
+        '403':
+          $ref: '#/components/responses/Forbidden'
+        '500':
+          $ref: '#/components/responses/InternalError'
+  /api/v1/vehicles/mileage/query:
+    post:
+      tags: [合作方数据接口]
+      summary: 查询车辆单日里程
+      operationId: queryDailyMileage
+      security:
+        - AppKeyAuth: []
+      requestBody:
+        required: true
+        content:
+          application/json:
+            schema:
+              $ref: '#/components/schemas/VehicleDailyQuery'
+            example:
+              plateNumbers: [粤A12345]
+              date: '2026-07-01'
+      responses:
+        '200':
+          description: 查询成功；无数据车辆仍保留在结果中
+          content:
+            application/json:
+              schema:
+                $ref: '#/components/schemas/MileageQueryResponse'
+        '400':
+          $ref: '#/components/responses/BadRequest'
+        '401':
+          $ref: '#/components/responses/Unauthorized'
+        '403':
+          $ref: '#/components/responses/Forbidden'
+        '500':
+          $ref: '#/components/responses/InternalError'
+  /api/v1/vehicles/total-mileage/query:
+    post:
+      tags: [合作方数据接口]
+      summary: 查询指定时刻的车辆总里程
+      description: |
+        返回不晚于请求时间的最近一条有效总里程记录、实际记录时间和时间差秒数。
+        protocol 不传时严格按 GB32960 > YUTONG_MQTT > JT808 选择首个有数据的协议。
+        GB32960 和 YUTONG_MQTT 为车辆仪表盘或车端累计里程；JT808 为定位终端/GPS侧累计里程。
+      operationId: queryTotalMileageAtTime
+      security:
+        - AppKeyAuth: []
+      requestBody:
+        required: true
+        content:
+          application/json:
+            schema:
+              $ref: '#/components/schemas/TotalMileageQuery'
+            example:
+              vin: LA9GG68L2PBAF4790
+              time: '2026-07-21 09:30:00'
+              protocol: '32960'
+      responses:
+        '200':
+          description: 查询成功；没有可用记录时返回 NO_DATA
+          content:
+            application/json:
+              schema:
+                $ref: '#/components/schemas/TotalMileageQueryResponse'
+        '400':
+          $ref: '#/components/responses/BadRequest'
+        '401':
+          $ref: '#/components/responses/Unauthorized'
+        '403':
+          $ref: '#/components/responses/Forbidden'
+        '500':
+          $ref: '#/components/responses/InternalError'
+  /api/v2/open-platform/apps:
+    get:
+      tags: [开放平台管理]
+      summary: 查询开放平台应用
+      operationId: listOpenPlatformApps
+      security:
+        - AdminBearer: []
+      responses:
+        '200':
+          description: 应用列表
+          content:
+            application/json:
+              schema:
+                $ref: '#/components/schemas/AdminAppListResponse'
+        '401':
+          description: 未登录
+        '403':
+          description: 非管理员
+    post:
+      tags: [开放平台管理]
+      summary: 创建应用和 appKey
+      description: appKey 明文仅在本次响应中返回。
+      operationId: createOpenPlatformApp
+      security:
+        - AdminBearer: []
+      requestBody:
+        required: true
+        content:
+          application/json:
+            schema:
+              $ref: '#/components/schemas/AppInput'
+      responses:
+        '200':
+          description: 创建成功
+          content:
+            application/json:
+              schema:
+                $ref: '#/components/schemas/AdminAppCreatedResponse'
+        '400':
+          description: 参数错误
+        '401':
+          description: 未登录
+        '403':
+          description: 非管理员
+  /api/v2/open-platform/apps/{id}:
+    put:
+      tags: [开放平台管理]
+      summary: 更新应用状态和有效期
+      operationId: updateOpenPlatformApp
+      security:
+        - AdminBearer: []
+      parameters:
+        - $ref: '#/components/parameters/AppId'
+      requestBody:
+        required: true
+        content:
+          application/json:
+            schema:
+              $ref: '#/components/schemas/AppInput'
+      responses:
+        '200':
+          description: 更新成功
+        '400':
+          description: 参数错误
+        '404':
+          description: 应用不存在
+  /api/v2/open-platform/apps/{id}/rotate-key:
+    post:
+      tags: [开放平台管理]
+      summary: 轮换 appKey
+      description: 旧 Key 立即失效，新 Key 明文仅在本次响应中返回。
+      operationId: rotateOpenPlatformAppKey
+      security:
+        - AdminBearer: []
+      parameters:
+        - $ref: '#/components/parameters/AppId'
+      responses:
+        '200':
+          description: 轮换成功
+          content:
+            application/json:
+              schema:
+                $ref: '#/components/schemas/AdminAppCreatedResponse'
+        '404':
+          description: 应用不存在
+  /api/v2/open-platform/apps/{id}/vehicles:
+    get:
+      tags: [开放平台管理]
+      summary: 查询应用的车辆授权
+      operationId: listOpenPlatformVehicleGrants
+      security:
+        - AdminBearer: []
+      parameters:
+        - $ref: '#/components/parameters/AppId'
+      responses:
+        '200':
+          description: 授权列表
+        '404':
+          description: 应用不存在
+    put:
+      tags: [开放平台管理]
+      summary: 完整替换车辆授权
+      operationId: replaceOpenPlatformVehicleGrants
+      security:
+        - AdminBearer: []
+      parameters:
+        - $ref: '#/components/parameters/AppId'
+      requestBody:
+        required: true
+        content:
+          application/json:
+            schema:
+              $ref: '#/components/schemas/VehicleGrantRequest'
+      responses:
+        '200':
+          description: 替换成功
+        '400':
+          description: VIN 或有效期不正确
+        '404':
+          description: 应用不存在
+components:
+  securitySchemes:
+    AppKeyAuth:
+      type: http
+      scheme: bearer
+      bearerFormat: 32-character appKey
+      description: 32 位无连字符 UUID appKey
+    AdminBearer:
+      type: http
+      scheme: bearer
+      bearerFormat: Platform session token
+      description: 车辆数据平台管理员令牌
+  parameters:
+    AppId:
+      name: id
+      in: path
+      required: true
+      schema:
+        type: integer
+        format: int64
+        minimum: 1
+  responses:
+    BadRequest:
+      description: 请求参数不正确
+      content:
+        application/json:
+          schema:
+            $ref: '#/components/schemas/ErrorResponse'
+          examples:
+            invalidDate:
+              value:
+                code: INVALID_DATE_FORMAT
+                message: date格式必须为yyyy-MM-dd
+                traceId: 4ccf63c4e51d4d4ab9107d931783a53e
+    Unauthorized:
+      description: appKey 不存在、停用或过期
+      content:
+        application/json:
+          schema:
+            $ref: '#/components/schemas/ErrorResponse'
+    Forbidden:
+      description: Key 或车辆授权未完整覆盖查询自然日
+      content:
+        application/json:
+          schema:
+            $ref: '#/components/schemas/ErrorResponse'
+    InternalError:
+      description: 服务内部异常
+      content:
+        application/json:
+          schema:
+            $ref: '#/components/schemas/ErrorResponse'
+  schemas:
+    VehicleDailyQuery:
+      type: object
+      additionalProperties: false
+      required: [plateNumbers, date]
+      properties:
+        plateNumbers:
+          type: array
+          minItems: 1
+          maxItems: 200
+          uniqueItems: true
+          items:
+            type: string
+            minLength: 1
+            maxLength: 32
+          description: 需要查询的车牌号
+        date:
+          type: string
+          format: date
+          description: 查询自然日，yyyy-MM-dd
+    TotalMileageQuery:
+      type: object
+      additionalProperties: false
+      required: [vin, time]
+      properties:
+        vin:
+          type: string
+          pattern: '^[A-HJ-NPR-Z0-9]{17}$'
+          description: 已授权车辆 VIN
+        time:
+          type: string
+          pattern: '^\\d{4}-\\d{2}-\\d{2} \\d{2}:\\d{2}:\\d{2}$'
+          description: 请求时刻，北京时间，固定格式 yyyy-MM-dd HH:mm:ss
+        protocol:
+          type: string
+          enum: ['32960', mqtt, '808', GB32960, YUTONG_MQTT, JT808]
+          description: 可选；不传时按 GB32960 > YUTONG_MQTT > JT808
+    HydrogenResult:
+      type: object
+      required: [plateNumber, date, hydrogenConsumptionKg, status]
+      properties:
+        plateNumber:
+          type: string
+        date:
+          type: string
+          format: date
+        hydrogenConsumptionKg:
+          type: number
+          format: double
+          nullable: true
+          description: 单日用氢量，kg；无数据时为 null
+        status:
+          $ref: '#/components/schemas/DataStatus'
+    MileageResult:
+      type: object
+      required: [plateNumber, date, dailyMileageKm, status]
+      properties:
+        plateNumber:
+          type: string
+        date:
+          type: string
+          format: date
+        dailyMileageKm:
+          type: number
+          format: double
+          nullable: true
+          description: 单日里程，km；无数据时为 null
+        status:
+          $ref: '#/components/schemas/DataStatus'
+    DataStatus:
+      type: string
+      enum: [NORMAL, NO_DATA]
+    HydrogenQueryResponse:
+      allOf:
+        - $ref: '#/components/schemas/SuccessEnvelope'
+        - type: object
+          required: [data]
+          properties:
+            data:
+              type: array
+              items:
+                $ref: '#/components/schemas/HydrogenResult'
+    MileageQueryResponse:
+      allOf:
+        - $ref: '#/components/schemas/SuccessEnvelope'
+        - type: object
+          required: [data]
+          properties:
+            data:
+              type: array
+              items:
+                $ref: '#/components/schemas/MileageResult'
+    TotalMileageResult:
+      type: object
+      required: [vin, queryTime, totalMileageKm, selectionPolicy, status]
+      properties:
+        vin:
+          type: string
+        queryTime:
+          type: string
+          description: 请求时刻，北京时间
+        totalMileageKm:
+          type: number
+          format: double
+          nullable: true
+          description: 总里程，km；无数据时为 null
+        protocol:
+          type: string
+          enum: [GB32960, YUTONG_MQTT, JT808]
+          description: 实际采用的采集协议
+        protocolInput:
+          type: string
+          description: 请求中指定的协议原值；未指定时省略
+        mileageMeaning:
+          type: string
+          description: 当前协议总里程的业务含义
+        recordTime:
+          type: string
+          description: 命中的实际采集记录时间，北京时间
+        timeDifferenceSeconds:
+          type: integer
+          format: int64
+          minimum: 0
+          description: 请求时间减实际记录时间，单位秒
+        selectionPolicy:
+          type: string
+          example: GB32960 > YUTONG_MQTT > JT808
+        status:
+          $ref: '#/components/schemas/DataStatus'
+    TotalMileageQueryResponse:
+      allOf:
+        - $ref: '#/components/schemas/SuccessEnvelope'
+        - type: object
+          required: [data]
+          properties:
+            data:
+              $ref: '#/components/schemas/TotalMileageResult'
+    SuccessEnvelope:
+      type: object
+      required: [code, message, traceId]
+      properties:
+        code:
+          type: string
+          enum: [SUCCESS]
+        message:
+          type: string
+          example: success
+        traceId:
+          type: string
+          example: 4ccf63c4e51d4d4ab9107d931783a53e
+    ErrorResponse:
+      type: object
+      required: [code, message, traceId]
+      properties:
+        code:
+          type: string
+          enum:
+            - INVALID_REQUEST
+            - PLATE_NUMBERS_EMPTY
+            - INVALID_DATE_FORMAT
+            - UNAUTHORIZED
+            - FORBIDDEN
+            - INTERNAL_ERROR
+        message:
+          type: string
+        traceId:
+          type: string
+    AppInput:
+      type: object
+      additionalProperties: false
+      required: [name, status, validFrom]
+      properties:
+        name:
+          type: string
+          maxLength: 96
+          example: 示例合作方
+        status:
+          type: string
+          enum: [enabled, disabled]
+        validFrom:
+          type: string
+          format: date-time
+          example: '2026-07-01T00:00:00+08:00'
+        validTo:
+          type: string
+          format: date-time
+          nullable: true
+          example: '2027-07-01T00:00:00+08:00'
+    App:
+      type: object
+      required: [id, name, appKeyPrefix, status, validFrom]
+      properties:
+        id:
+          type: integer
+          format: int64
+        name:
+          type: string
+        appKeyPrefix:
+          type: string
+          minLength: 8
+          maxLength: 8
+        status:
+          type: string
+          enum: [enabled, disabled]
+        validFrom:
+          type: string
+          format: date-time
+        validTo:
+          type: string
+          format: date-time
+          nullable: true
+    AppCreated:
+      allOf:
+        - $ref: '#/components/schemas/App'
+        - type: object
+          required: [appKey]
+          properties:
+            appKey:
+              type: string
+              minLength: 32
+              maxLength: 32
+              pattern: '^[0-9a-f]{32}$'
+              description: 仅本次响应返回
+    VehicleGrantRequest:
+      type: object
+      additionalProperties: false
+      required: [vehicles]
+      properties:
+        vehicles:
+          type: array
+          maxItems: 500
+          items:
+            $ref: '#/components/schemas/VehicleGrantInput'
+    VehicleGrantInput:
+      type: object
+      additionalProperties: false
+      required: [vin, validFrom]
+      properties:
+        vin:
+          type: string
+          minLength: 17
+          maxLength: 17
+        validFrom:
+          type: string
+          format: date-time
+        validTo:
+          type: string
+          format: date-time
+          nullable: true
+    AdminAppListResponse:
+      type: object
+      properties:
+        data:
+          type: array
+          items:
+            $ref: '#/components/schemas/App'
+        traceId:
+          type: string
+    AdminAppCreatedResponse:
+      type: object
+      properties:
+        data:
+          $ref: '#/components/schemas/AppCreated'
+        traceId:
+          type: string
diff --git a/post_process_transfer.py b/post_process_transfer.py
new file mode 100644
index 0000000..8fb8835
--- /dev/null
+++ b/post_process_transfer.py
@@ -0,0 +1,235 @@
+#!/usr/bin/env python3
+"""后处理：在核算文件中体现离职转嫁 — 当月转新销售，其他不发了"""
+import openpyxl
+from openpyxl.styles import Font, PatternFill, Alignment, Border, Side
+from copy import copy
+import os
+
+os.chdir('/Users/kkfluous/Downloads/20260713')
+fp = '里程任务考核_6月核算.xlsx'
+
+wb = openpyxl.load_workbook(fp)
+ws = wb['6月汇总']
+
+# ── Styles ──
+HEAD_FILL = PatternFill('solid', fgColor='4472C4')
+HEAD_FONT = Font(bold=True, color='FFFFFF', size=10, name='宋体')
+SUB_FILL  = PatternFill('solid', fgColor='D9E1F2')
+TITLE_FNT = Font(bold=True, size=12, name='宋体')
+NOTE_FNT  = Font(size=10, color='C00000', name='宋体')
+NORM_FNT  = Font(size=10, name='宋体')
+BOLD_FNT  = Font(bold=True, size=10, name='宋体')
+THIN = Side('thin', color='B4B4B4')
+BORD = Border(THIN, THIN, THIN, THIN)
+CC = Alignment(horizontal='center', vertical='center', wrap_text=True)
+LL = Alignment(horizontal='left', vertical='center', wrap_text=True)
+YELLOW_FILL = PatternFill('solid', fgColor='FFF2CC')
+
+# ── Find key rows ──
+row_考核应发合计 = None
+row_最终发放_title = None
+row_最终发放明细 = None
+row_按销售人员 = None
+row_按部门 = None
+row_总览 = None
+
+for r in range(1, ws.max_row + 1):
+    v = str(ws.cell(row=r, column=1).value or '')
+    if '考核应发合计' in v and r > 50:
+        row_考核应发合计 = r
+    elif '三、最终发放' in v:
+        row_最终发放_title = r
+    elif v == '最终发放明细':
+        row_最终发放明细 = r
+    elif '6月最终发放（按销售人员）' in v:
+        row_按销售人员 = r
+    elif '6月最终发放（按部门）' in v:
+        row_按部门 = r
+    elif v == '总览':
+        row_总览 = r
+
+print(f"考核应发合计: row {row_考核应发合计}")
+print(f"三、最终发放: row {row_最终发放_title}")
+print(f"最终发放明细: row {row_最终发放明细}")
+print(f"按销售人员: row {row_按销售人员}")
+print(f"按部门: row {row_按部门}")
+print(f"总览: row {row_总览}")
+
+# ── Transfer config ──
+# (离职人员, 接收人, 当月转嫁金额, 不发了金额)
+TRANSFERS = {
+    '赵连飞': {'to': '刘念念', 'transfer': 2260.00, 'dept': '业务二部'},
+    '伍仲文': {'to': '钟祥',   'transfer': 150.00,  'dept': '业务六部'},
+    '岑彦':   {'to': '钟祥',   'transfer': 30.00,   'dept': '业务六部'},
+}
+
+# ── Step 1: Insert "二、离职人员当月奖金转嫁" section ──
+# Insert 9 rows after 考核应发合计 row
+insert_at = row_考核应发合计 + 1
+ws.insert_rows(insert_at, 9)
+
+# Write transfer section
+r = insert_at
+ws.merge_cells(start_row=r, start_column=1, end_row=r, end_column=3)
+ws.cell(row=r, column=1, value='二、离职人员当月奖金转嫁').font = TITLE_FNT
+r += 2
+
+for c, h in enumerate(['离职人员', '转嫁至', '转嫁金额', '说明'], 1):
+    cell = ws.cell(row=r, column=c, value=h)
+    cell.font = HEAD_FONT; cell.fill = HEAD_FILL; cell.alignment = CC; cell.border = BORD
+r += 1
+
+total_transfer = 0
+for name, info in TRANSFERS.items():
+    ws.cell(row=r, column=1, value=name).font = NORM_FNT
+    ws.cell(row=r, column=1).alignment = CC; ws.cell(row=r, column=1).border = BORD
+    ws.cell(row=r, column=2, value=info['to']).font = NORM_FNT
+    ws.cell(row=r, column=2).alignment = CC; ws.cell(row=r, column=2).border = BORD
+    ws.cell(row=r, column=3, value=info['transfer']).font = NORM_FNT
+    ws.cell(row=r, column=3).alignment = CC; ws.cell(row=r, column=3).border = BORD
+    ws.cell(row=r, column=3).number_format = '#,##0.00'
+    ws.cell(row=r, column=4, value='当月达标奖金转新销售（需确认是否发放）').font = NORM_FNT
+    ws.cell(row=r, column=4).alignment = LL; ws.cell(row=r, column=4).border = BORD
+    total_transfer += info['transfer']
+    r += 1
+
+# 合计行
+for c in range(1, 5):
+    ws.cell(row=r, column=c).fill = SUB_FILL; ws.cell(row=r, column=c).border = BORD
+    ws.cell(row=r, column=c).font = BOLD_FNT; ws.cell(row=r, column=c).alignment = CC
+ws.cell(row=r, column=1, value='合计')
+ws.cell(row=r, column=3, value=total_transfer)
+ws.cell(row=r, column=3).number_format = '#,##0.00'
+
+# ── Step 2: Update 最终发放明细 (shifted down by 9 rows) ──
+# Find the shifted positions
+shift = 9
+row_detail_header = row_最终发放明细 + shift  # 最终发放明细 header
+row_detail_data_start = row_detail_header + 2  # first data row (after "销售人员|车辆数|金额")
+
+print(f"\n最终发放明细 shifted to header row {row_detail_header}, data starts at {row_detail_data_start}")
+
+# Parse and update the data rows
+# Data rows are from row_detail_data_start to the 总计 row
+resigned_names = set(TRANSFERS.keys())
+receiver_additions = {'刘念念': 2260.0, '钟祥': 180.0}
+
+for r in range(row_detail_data_start, ws.max_row + 1):
+    name_cell = ws.cell(row=r, column=1)
+    amt_cell = ws.cell(row=r, column=3)
+    name = str(name_cell.value or '').strip()
+
+    if name == '总计':
+        # Update total (should remain same: 113408.28)
+        break
+
+    if not name or name == '':
+        break
+
+    # Update receiver amounts
+    if name in receiver_additions:
+        old_val = float(amt_cell.value or 0)
+        new_val = old_val + receiver_additions[name]
+        amt_cell.value = new_val
+        print(f"  {name}: {old_val:.2f} → {new_val:.2f} (+{receiver_additions[name]:.2f})")
+
+    # Update resigned amounts (keep only不发了)
+    if name in TRANSFERS:
+        old_val = float(amt_cell.value or 0)
+        new_val = old_val - TRANSFERS[name]['transfer']
+        amt_cell.value = new_val
+        # Add note
+        note_cell = ws.cell(row=r, column=4)
+        note_cell.value = '不发了（仅当月转嫁）'
+        note_cell.font = NOTE_FNT
+        note_cell.alignment = CC
+        print(f"  {name}: {old_val:.2f} → {new_val:.2f} (不发了)")
+
+# ── Step 3: Update 按销售人员 section ──
+row_sales_header = row_按销售人员 + shift
+print(f"\n按销售人员 shifted to header row {row_sales_header}")
+
+for r in range(row_sales_header + 1, ws.max_row + 1):
+    name_cell = ws.cell(row=r, column=1)
+    amt_cell = ws.cell(row=r, column=3)
+    name = str(name_cell.value or '').strip()
+
+    if name == '合计':
+        break
+
+    if not name:
+        break
+
+    if name in receiver_additions:
+        old_val = float(amt_cell.value or 0)
+        amt_cell.value = old_val + receiver_additions[name]
+
+    if name in TRANSFERS:
+        old_val = float(amt_cell.value or 0)
+        amt_cell.value = old_val - TRANSFERS[name]['transfer']
+
+# ── Step 4: Update 总览 section (add notes) ──
+row_overview = row_总览 + shift
+print(f"\n总览 shifted to row {row_overview}")
+
+# Add notes below总览
+note_row = row_overview + 5  # after "最终发放" row
+ws.merge_cells(start_row=note_row, start_column=1, end_row=note_row, end_column=3)
+ws.cell(row=note_row, column=1, value='').font = NORM_FNT
+note_row += 1
+
+ws.merge_cells(start_row=note_row, start_column=1, end_row=note_row, end_column=3)
+ws.cell(row=note_row, column=1, value='备注：').font = BOLD_FNT
+note_row += 1
+
+notes = [
+    '1. 离职人员（赵连飞、伍仲文、岑彦）当月达标奖金转嫁至接收人；结转、补发、累计补发等一律不发放。',
+    '2. 转嫁金额需与相关部门确认是否发放给新销售。',
+    '3. 赵连飞（2026-04-30离职）→ 刘念念；伍仲文（2026-05-09离职）→ 钟祥；岑彦（2026-05-09离职）→ 钟祥。',
+]
+for note in notes:
+    ws.merge_cells(start_row=note_row, start_column=1, end_row=note_row, end_column=3)
+    ws.cell(row=note_row, column=1, value=note).font = NOTE_FNT
+    ws.cell(row=note_row, column=1).alignment = LL
+    note_row += 1
+
+# ── Step 5: Rename resigned sheets with 【已离职】prefix and move to end ──
+resigned_sheet_names = {
+    '二部-赵连飞': '【已离职】二部-赵连飞',
+    '六部-伍仲文': '【已离职】六部-伍仲文',
+    '六部-岑彦':   '【已离职】六部-岑彦',
+}
+
+for old_name, new_name in resigned_sheet_names.items():
+    if old_name in wb.sheetnames:
+        idx = wb.sheetnames.index(old_name)
+        wb[old_name].title = new_name
+        # Move to end
+        wb.move_sheet(new_name, offset=len(wb.sheetnames) - 1 - idx)
+        print(f"  Sheet renamed: {old_name} → {new_name} (moved to end)")
+
+# ── Save ──
+wb.save(fp)
+print(f"\n✅ {fp} 已更新")
+
+# ── Verify ──
+wb2 = openpyxl.load_workbook(fp, data_only=True)
+ws2 = wb2['6月汇总']
+print("\n验证 - 最终发放明细:")
+in_detail = False
+for r in range(1, ws2.max_row + 1):
+    v = str(ws2.cell(row=r, column=1).value or '')
+    if v == '最终发放明细':
+        in_detail = True
+        continue
+    if in_detail:
+        name = ws2.cell(row=r, column=1).value
+        amt = ws2.cell(row=r, column=3).value
+        note = ws2.cell(row=r, column=4).value
+        if name and name != '销售人员':
+            print(f"  {name}: {amt}" + (f" [{note}]" if note else ""))
+        if name == '总计':
+            break
+
+print("\nSheets:", wb2.sheetnames)
+wb2.close()
diff --git a/prepare_july_source.py b/prepare_july_source.py
new file mode 100644
index 0000000..3b77e4f
--- /dev/null
+++ b/prepare_july_source.py
@@ -0,0 +1,167 @@
+#!/usr/bin/env python3
+"""按三份GPS日里程文件处理2026年7月测试车辆考核源表。"""
+from copy import copy
+from collections import defaultdict
+from datetime import datetime
+from pathlib import Path
+import re
+import shutil
+import openpyxl
+from openpyxl.styles import PatternFill
+
+BASE = Path('/Users/kkfluous/Downloads')
+OUT = BASE / '7月核算结果'
+SRC = BASE / '租赁任务考核_2026年7月 (1).xlsx'
+TEST = Path('/Users/kkfluous/Library/Containers/com.tencent.xinWeChat/Data/Documents/xwechat_files/wxid_1704407055112_3af8/temp/RWTemp/2026-08/4340099467fbf08b4c6583c49adba4ee/7月测试车辆明细.xlsx')
+GPS_DAILY = BASE / '智能管车_车辆里程日报-140106360.xlsx'
+GPS_DETAIL = [
+    BASE / '里程统计[天][2026-07-01至2026-07-31].xlsx',
+    BASE / '里程统计[天][2026-07-01至2026-07-31] (1).xlsx',
+]
+EXCLUDED_TEST_PLATES = {'粤AG37316'}
+ZERO_OVERRIDE = set()
+DEPARTMENT_OVERRIDE = {'粤AGP6675': '业务六部', '粤AGP9757': '业务六部'}
+
+def norm(v):
+    return re.sub(r'\s+', '', str(v or '')).upper()
+
+def number(v):
+    try:
+        return float(v or 0)
+    except (TypeError, ValueError):
+        return 0.0
+
+def read_test_plates():
+    wb = openpyxl.load_workbook(TEST, data_only=True)
+    ws = wb.active
+    plates = {norm(r[0]) for r in ws.iter_rows(min_row=2, values_only=True) if norm(r[0])}
+    wb.close()
+    return plates - EXCLUDED_TEST_PLATES
+
+def read_daily_matrix():
+    wb = openpyxl.load_workbook(GPS_DAILY, data_only=True)
+    ws = wb.active
+    headers = [c.value for c in ws[1]]
+    day_cols = {}
+    for i, v in enumerate(headers):
+        m = re.fullmatch(r'07月(\d{2})日', str(v or ''))
+        if m:
+            day_cols[int(m.group(1))] = i
+    data = {}
+    for row in ws.iter_rows(min_row=2, values_only=True):
+        plate = norm(row[0])
+        if plate:
+            data[plate] = {d: number(row[i]) for d, i in day_cols.items()}
+    wb.close()
+    return data
+
+def read_detail_files():
+    data = defaultdict(lambda: defaultdict(float))
+    for path in GPS_DETAIL:
+        wb = openpyxl.load_workbook(path, data_only=True)
+        ws = wb.active
+        headers = [c.value for c in ws[3]]
+        pcol = next(i for i, v in enumerate(headers) if v and '车牌' in str(v))
+        dcol = next(i for i, v in enumerate(headers) if str(v or '').strip() == '日期')
+        mcol = next(i for i, v in enumerate(headers) if v and '里程' in str(v) and 'km' in str(v).lower())
+        for row in ws.iter_rows(min_row=4, values_only=True):
+            plate = norm(row[pcol])
+            if not plate:
+                continue
+            try:
+                dt = row[dcol] if isinstance(row[dcol], datetime) else datetime.strptime(str(row[dcol])[:10], '%Y-%m-%d')
+            except (TypeError, ValueError):
+                continue
+            # 同一车辆同一天在两份明细中如重复，取较大值，避免重复累计。
+            data[plate][dt.day] = max(data[plate][dt.day], number(row[mcol]))
+        wb.close()
+    return {p: dict(days) for p, days in data.items()}
+
+def copy_cell_style(src, dst):
+    dst._style = copy(src._style)
+    if src.has_style:
+        dst.number_format = src.number_format
+        dst.protection = copy(src.protection)
+        dst.alignment = copy(src.alignment)
+
+def main():
+    OUT.mkdir(parents=True, exist_ok=True)
+    shutil.copy2(SRC, OUT / '租赁任务考核_2026年7月_原版备份.xlsx')
+    tests = read_test_plates()
+    matrix = read_daily_matrix()
+    details = read_detail_files()
+
+    wb = openpyxl.load_workbook(SRC)
+    ws = wb['业务考核视图']
+    # 原表O列是实际里程。插入后统一为 O原TBOX、P实际、Q测试、R完成率、S是否达标。
+    ws.insert_cols(15, 1)
+    ws.insert_cols(17, 1)
+    ws['O1'] = '原TBOX里程(km)'
+    ws['Q1'] = '测试里程(km)'
+    copy_cell_style(ws['P1'], ws['O1'])
+    copy_cell_style(ws['P1'], ws['Q1'])
+    yellow = PatternFill(fill_type='solid', fgColor='FFFF00')
+    records = []
+    missing = []
+    for ri in range(2, ws.max_row + 1):
+        plate = norm(ws.cell(ri, 1).value)
+        if plate in DEPARTMENT_OVERRIDE:
+            ws.cell(ri, 4).value = DEPARTMENT_OVERRIDE[plate]
+        original = number(ws.cell(ri, 16).value)
+        ws.cell(ri, 15).value = round(original, 2)
+        copy_cell_style(ws.cell(ri, 16), ws.cell(ri, 15))
+        copy_cell_style(ws.cell(ri, 16), ws.cell(ri, 17))
+        if plate not in tests:
+            ws.cell(ri, 17).value = 0
+            continue
+        start = ws.cell(ri, 11).value
+        end = ws.cell(ri, 12).value
+        start_day = start.day if isinstance(start, datetime) else 1
+        end_day = end.day if isinstance(end, datetime) else 31
+        if plate in ZERO_OVERRIDE:
+            actual = 0.0
+            source = 'GPS缺失，业务确认暂按0'
+        elif plate in matrix:
+            actual = sum(matrix[plate].get(d, 0) for d in range(start_day, end_day + 1))
+            source = GPS_DAILY.name
+        elif plate in details:
+            actual = sum(details[plate].get(d, 0) for d in range(start_day, end_day + 1))
+            source = '里程统计日明细'
+        else:
+            missing.append(plate)
+            continue
+        target = number(ws.cell(ri, 14).value)
+        test_mileage = max(0.0, original - actual)
+        rate = actual / target * 100 if target > 0 else 0
+        ws.cell(ri, 16).value = round(actual, 2)
+        ws.cell(ri, 17).value = round(test_mileage, 2)
+        ws.cell(ri, 18).value = round(rate, 2)
+        ws.cell(ri, 19).value = '达标' if target > 0 and actual >= target else '未达标'
+        for cell in ws[ri]:
+            cell.fill = copy(yellow)
+        records.append((plate, ri, original, actual, test_mileage, source))
+    if missing:
+        raise RuntimeError('仍缺GPS车牌: ' + ','.join(sorted(set(missing))))
+
+    # 对齐5/6月的宽度、筛选和冻结窗格；保留原表其余样式。
+    ref = openpyxl.load_workbook(BASE / '租赁任务考核_2026年6月.xlsx', read_only=False)
+    rws = ref['业务考核视图']
+    for i in range(1, 24):
+        letter = openpyxl.utils.get_column_letter(i)
+        if rws.column_dimensions[letter].width:
+            ws.column_dimensions[letter].width = rws.column_dimensions[letter].width
+    ws.freeze_panes = rws.freeze_panes
+    ws.auto_filter.ref = f'A1:W{ws.max_row}'
+    ref.close()
+
+    out = OUT / '租赁任务考核_2026年7月.xlsx'
+    wb.save(out)
+    wb.close()
+    print(f'output={out}')
+    print(f'test_list={len(tests)} intervention_records={len(records)} intervention_plates={len(set(r[0] for r in records))}')
+    print(f'zero_override={sorted(ZERO_OVERRIDE)}')
+    print(f'department_override={DEPARTMENT_OVERRIDE}')
+    print(f'total_original={sum(r[2] for r in records):.2f} total_gps={sum(r[3] for r in records):.2f} total_test={sum(r[4] for r in records):.2f}')
+
+if __name__ == '__main__':
+    main()
diff --git a/rebuild_july_source_from_multi_source.py b/rebuild_july_source_from_multi_source.py
new file mode 100644
index 0000000..14ec26d
--- /dev/null
+++ b/rebuild_july_source_from_multi_source.py
@@ -0,0 +1,75 @@
+#!/usr/bin/env python3
+"""将用户确认的多源核算里程按车牌+考核区间回填到7月考核源表。"""
+import os
+import re
+import shutil
+from datetime import datetime
+from pathlib import Path
+
+import openpyxl
+
+BASE = Path('/Users/kkfluous/Downloads')
+MILEAGE = BASE / '7月考核车辆多源里程汇总_用户确认版.xlsx'
+SOURCE = BASE / '7月核算结果/租赁任务考核_2026年7月.xlsx'
+OUTPUT_DIR = Path(os.environ.get('JULY_OUTPUT_DIR', BASE / '7月绩效重算_多源考核里程_20260817'))
+OUTPUT = OUTPUT_DIR / '租赁任务考核_2026年7月.xlsx'
+
+def norm(v):
+    return re.sub(r'\s+', '', str(v or '')).upper()
+
+def day(v):
+    return v.date() if isinstance(v, datetime) else v
+
+OUTPUT_DIR.mkdir(parents=True, exist_ok=True)
+shutil.copy2(SOURCE, OUTPUT)
+
+mw = openpyxl.load_workbook(MILEAGE, data_only=True)
+ms = mw['7月考核车辆里程汇总']
+mh = {ms.cell(2, c).value: c for c in range(1, ms.max_column + 1)}
+mileage = {}
+for r in range(3, ms.max_row + 1):
+    key = (norm(ms.cell(r, mh['车牌']).value), day(ms.cell(r, mh['考核开始日期']).value),
+           day(ms.cell(r, mh['考核结束日期']).value))
+    value = ms.cell(r, mh['核算里程(km)']).value
+    if key in mileage:
+        raise RuntimeError(f'多源汇总存在重复键: {key}')
+    if not isinstance(value, (int, float)):
+        raise RuntimeError(f'核算里程不是数值: {key}={value}')
+    mileage[key] = float(value)
+mw.close()
+
+wb = openpyxl.load_workbook(OUTPUT)
+ws = wb['业务考核视图']
+h = {ws.cell(1, c).value: c for c in range(1, ws.max_column + 1)}
+used = set()
+changed = 0
+for r in range(2, ws.max_row + 1):
+    key = (norm(ws.cell(r, h['车牌号']).value), day(ws.cell(r, h['考核开始日期']).value),
+           day(ws.cell(r, h['考核结束日期']).value))
+    if key not in mileage:
+        raise RuntimeError(f'7月源表记录在多源汇总中缺失: {key}')
+    used.add(key)
+    actual = round(mileage[key], 2)
+    original = float(ws.cell(r, h['原TBOX里程(km)']).value or 0)
+    target = float(ws.cell(r, h['应考核里程(km)']).value or 0)
+    old_actual = float(ws.cell(r, h['实际行驶里程(km)']).value or 0)
+    if abs(old_actual - actual) > 0.005:
+        changed += 1
+    ws.cell(r, h['实际行驶里程(km)']).value = actual
+    ws.cell(r, h['测试里程(km)']).value = round(max(0, original - actual), 2)
+    rate = round(actual / target * 100, 2) if target > 0 else 0
+    ws.cell(r, h['完成率(%)']).value = rate
+    ws.cell(r, h['是否达标']).value = '达标' if actual >= target and target > 0 else '未达标'
+    for name in ['原TBOX里程(km)', '实际行驶里程(km)', '测试里程(km)']:
+        ws.cell(r, h[name]).number_format = '#,##0.00'
+    ws.cell(r, h['完成率(%)']).number_format = '0.00'
+
+extra = set(mileage) - used
+if extra:
+    raise RuntimeError(f'多源汇总存在源表未使用记录: {list(extra)[:5]}')
+
+ws.freeze_panes = 'A2'
+wb.save(OUTPUT)
+wb.close()
+print({'output': str(OUTPUT), 'rows': len(used), 'changed_rows': changed})
+
diff --git a/rebuild_summary.py b/rebuild_summary.py
new file mode 100644
index 0000000..89a109d
--- /dev/null
+++ b/rebuild_summary.py
@@ -0,0 +1,315 @@
+#!/usr/bin/env python3
+"""按参考格式重建 6月汇总 sheet：剔除离职，转嫁用接收人格式"""
+import openpyxl
+from openpyxl.styles import Font, PatternFill, Alignment, Border, Side
+from openpyxl.utils import get_column_letter
+from collections import defaultdict, OrderedDict
+import os, copy
+
+os.chdir('/Users/kkfluous/Downloads')
+fp = '里程任务考核_6月核算.xlsx'
+
+# ── Config ──
+RESIGNED = {'赵连飞', '伍仲文', '岑彦'}
+TRANSFERS = [
+    ('刘念念', '赵连飞', 2260.00),
+    ('钟祥',   '伍仲文、岑彦', 180.00),
+]
+TRANSFER_TOTAL = sum(t[2] for t in TRANSFERS)
+# Map resigned -> receiver for aggregation
+TRANSFER_MAP = {'赵连飞': ('刘念念', 2260.00), '伍仲文': ('钟祥', 150.00), '岑彦': ('钟祥', 30.00)}
+
+# ── Styles ──
+HEAD_FILL = PatternFill('solid', fgColor='4472C4')
+HEAD_FONT = Font(bold=True, color='FFFFFF', size=10, name='宋体')
+SUB_FILL  = PatternFill('solid', fgColor='D9E1F2')
+TITLE_FNT = Font(bold=True, size=12, name='宋体')
+SECTION_FNT = Font(bold=True, size=11, name='宋体')
+NOTE_FNT  = Font(size=10, color='C00000', name='宋体')
+NORM_FNT  = Font(size=10, name='宋体')
+BOLD_FNT  = Font(bold=True, size=10, name='宋体')
+THIN = Side('thin', color='B4B4B4')
+BORD = Border(THIN, THIN, THIN, THIN)
+CC = Alignment(horizontal='center', vertical='center', wrap_text=True)
+LL = Alignment(horizontal='left', vertical='center', wrap_text=True)
+
+def hdr_row(ws, r, headers):
+    for c, h in enumerate(headers, 1):
+        cell = ws.cell(row=r, column=c, value=h)
+        cell.font = HEAD_FONT; cell.fill = HEAD_FILL; cell.alignment = CC; cell.border = BORD
+
+def data_row(ws, r, values, bold=False):
+    for c, v in enumerate(values, 1):
+        cell = ws.cell(row=r, column=c, value=v)
+        cell.font = BOLD_FNT if bold else NORM_FNT
+        cell.alignment = CC; cell.border = BORD
+        if isinstance(v, float):
+            cell.number_format = '#,##0.00'
+
+def sub_total(ws, r, values):
+    for c, v in enumerate(values, 1):
+        cell = ws.cell(row=r, column=c, value=v)
+        cell.font = BOLD_FNT; cell.fill = SUB_FILL
+        cell.alignment = CC; cell.border = BORD
+        if isinstance(v, float):
+            cell.number_format = '#,##0.00'
+
+# ── Step 1: Parse existing data ──
+wb = openpyxl.load_workbook(fp)
+ws_old = wb['6月汇总']
+
+# Parse all考核应发 subsections
+subsections = []  # [(title, [(name, vehicles, amount), ...], total_row)]
+current_title = None
+current_rows = []
+in_detail = False
+
+# Map person->dept from按销售人员 section
+person_dept = {}
+in_sales = False
+for r in range(1, ws_old.max_row + 1):
+    v1 = str(ws_old.cell(row=r, column=1).value or '')
+    v2 = str(ws_old.cell(row=r, column=2).value or '')
+    if '6月最终发放（按销售人员）' in v1:
+        in_sales = True
+        continue
+    if in_sales:
+        name = str(ws_old.cell(row=r, column=1).value or '')
+        dept = str(ws_old.cell(row=r, column=2).value or '')
+        if name == '合计':
+            break
+        if name and name != '销售人员' and dept:
+            person_dept[name] = dept
+
+print(f"人员部门映射: {person_dept}")
+
+# Parse考核应发 subsections
+for r in range(1, ws_old.max_row + 1):
+    v1 = str(ws_old.cell(row=r, column=1).value or '')
+
+    if v1.startswith('考核应发-'):
+        if current_title and current_rows:
+            subsections.append((current_title, current_rows))
+        current_title = v1
+        current_rows = []
+        in_detail = True
+        continue
+
+    if current_title and in_detail:
+        name = str(ws_old.cell(row=r, column=1).value or '')
+        vehicles = ws_old.cell(row=r, column=2).value
+        amount = ws_old.cell(row=r, column=3).value
+
+        if name == '总计':
+            subsections.append((current_title, current_rows))
+            current_title = None
+            current_rows = []
+            in_detail = False
+        elif name and name != '销售人员':
+            current_rows.append((name, vehicles, amount))
+
+print(f"\n解析到 {len(subsections)} 个考核应发 subsection:")
+
+# ── Step 2: Filter out resigned, recalculate ──
+filtered_subsections = []
+grand_total = 0.0
+
+for title, rows in subsections:
+    filtered = [(n, v, a) for (n, v, a) in rows if n not in RESIGNED]
+    # Recalculate vehicles and amounts
+    new_vehicles = sum(int(v or 0) for _, v, _ in filtered)
+    new_amount = sum(float(a or 0) for _, _, a in filtered)
+    grand_total += new_amount
+    filtered_subsections.append((title, filtered, new_vehicles, new_amount))
+    removed = [(n, a) for (n, _, a) in rows if n in RESIGNED]
+    if removed:
+        print(f"  {title}: 剔除 {removed}, 新合计={new_amount:.2f}")
+
+print(f"\n考核应发合计: {grand_total:.2f}")
+
+# ── Step 3: Build最终发放 (aggregate by person + transfers) ──
+person_data = defaultdict(lambda: {'vehicles': 0, 'amount': 0.0})
+for title, rows, _, _ in filtered_subsections:
+    for name, vehicles, amount in rows:
+        person_data[name]['vehicles'] += int(vehicles or 0)
+        person_data[name]['amount'] += float(amount or 0)
+
+# Add transfers to receivers
+for resigned, (receiver, amt) in TRANSFER_MAP.items():
+    person_data[receiver]['amount'] += amt
+    # Count vehicles? The reference doesn't seem to add vehicle count for transfers
+
+# Remove resigned (should already be filtered, but just in case)
+for name in list(person_data.keys()):
+    if name in RESIGNED:
+        del person_data[name]
+
+final_total = sum(d['amount'] for d in person_data.values())
+print(f"最终发放合计: {final_total:.2f}")
+
+# Sort by department then name (matching reference order)
+dept_order = ['业务二部', '业务三部', '业务五部', '业务六部']
+def sort_key(item):
+    name = item[0]
+    dept = person_dept.get(name, '')
+    try:
+        di = dept_order.index(dept)
+    except ValueError:
+        di = 99
+    return (di, name)
+
+sorted_persons = sorted(person_data.items(), key=sort_key)
+
+# ── Step 4: Department totals ──
+dept_totals = defaultdict(float)
+for name, data in sorted_persons:
+    dept = person_dept.get(name, '')
+    dept_totals[dept] += data['amount']
+
+# ── Step 5: Rebuild the sheet ──
+# Delete old sheet and create new
+ws = wb['6月汇总']
+ws.delete_rows(1, ws.max_row)
+
+rn = 1
+
+# === 一、考核应发明细 ===
+ws.merge_cells(start_row=rn, start_column=1, end_row=rn, end_column=3)
+ws.cell(row=rn, column=1, value='一、考核应发明细').font = TITLE_FNT
+rn += 2
+
+for title, rows, total_veh, total_amt in filtered_subsections:
+    ws.cell(row=rn, column=1, value=title).font = SECTION_FNT
+    rn += 1
+    hdr_row(ws, rn, ['销售人员', '车辆数', '金额'])
+    rn += 1
+
+    # Sort rows by dept then name
+    sorted_rows = sorted(rows, key=lambda x: sort_key((x[0],)) if x[0] in person_dept else (99, x[0]))
+
+    for name, vehicles, amount in sorted_rows:
+        data_row(ws, rn, [name, vehicles, round(float(amount or 0), 2)])
+        rn += 1
+
+    sub_total(ws, rn, ['总计', total_veh, round(total_amt, 2)])
+    rn += 2
+
+# 考核应发合计
+ws.merge_cells(start_row=rn, start_column=1, end_row=rn, end_column=2)
+ws.cell(row=rn, column=1, value='考核应发合计').font = BOLD_FNT
+ws.cell(row=rn, column=1).alignment = CC
+ws.cell(row=rn, column=3, value=round(grand_total, 2)).font = BOLD_FNT
+ws.cell(row=rn, column=3).alignment = CC
+ws.cell(row=rn, column=3).number_format = '#,##0.00'
+for c in range(1, 4):
+    ws.cell(row=rn, column=c).border = BORD
+    ws.cell(row=rn, column=c).fill = SUB_FILL
+rn += 2
+
+# === 二、离职人员转嫁收入 ===
+ws.merge_cells(start_row=rn, start_column=1, end_row=rn, end_column=3)
+ws.cell(row=rn, column=1, value='二、离职人员转嫁收入').font = TITLE_FNT
+rn += 1
+hdr_row(ws, rn, ['接收人', '来源', '转嫁金额'])
+rn += 1
+
+for receiver, source, amt in TRANSFERS:
+    data_row(ws, rn, [receiver, source, amt])
+    rn += 1
+
+sub_total(ws, rn, ['合计', '', TRANSFER_TOTAL])
+rn += 2
+
+# === 三、最终发放 ===
+ws.merge_cells(start_row=rn, start_column=1, end_row=rn, end_column=3)
+ws.cell(row=rn, column=1, value='三、最终发放').font = TITLE_FNT
+rn += 2
+
+# 最终发放明细
+ws.cell(row=rn, column=1, value='最终发放明细').font = SECTION_FNT
+rn += 1
+hdr_row(ws, rn, ['销售人员', '车辆数', '金额'])
+rn += 1
+
+for name, data in sorted_persons:
+    data_row(ws, rn, [name, data['vehicles'], round(data['amount'], 2)])
+    rn += 1
+
+sub_total(ws, rn, ['总计', sum(d['vehicles'] for _, d in sorted_persons), round(final_total, 2)])
+rn += 2
+
+# 6月最终发放（按销售人员）
+ws.cell(row=rn, column=1, value='6月最终发放（按销售人员）').font = SECTION_FNT
+rn += 1
+hdr_row(ws, rn, ['销售人员', '部门名称', '最终发放'])
+rn += 1
+
+for name, data in sorted_persons:
+    dept = person_dept.get(name, '')
+    data_row(ws, rn, [name, dept, round(data['amount'], 2)])
+    rn += 1
+
+sub_total(ws, rn, ['合计', '', round(final_total, 2)])
+rn += 2
+
+# 6月最终发放（按部门）
+ws.cell(row=rn, column=1, value='6月最终发放（按部门）').font = SECTION_FNT
+rn += 1
+hdr_row(ws, rn, ['部门名称', '最终发放', ''])
+rn += 1
+
+for dept in dept_order:
+    if dept in dept_totals:
+        data_row(ws, rn, [dept, round(dept_totals[dept], 2), ''])
+        rn += 1
+
+sub_total(ws, rn, ['合计', round(sum(dept_totals.values()), 2), ''])
+rn += 2
+
+# === 总览 ===
+ws.cell(row=rn, column=1, value='总览').font = SECTION_FNT
+rn += 1
+hdr_row(ws, rn, ['', '', ''])
+# Override header style for总览
+data_row(ws, rn, ['考核应发', round(final_total, 2), ''])
+rn += 1
+data_row(ws, rn, ['亏损拦截', 0, ''])
+rn += 1
+data_row(ws, rn, ['最终发放', round(final_total, 2), ''])
+rn += 2
+
+# === 备注 ===
+note_text = '备注：赵连飞（2026-04-30离职）、伍仲文（2026-05-09离职）、岑彦（2026-05-09离职）\n三人已从上述汇总中剔除，其6月当月达标奖金已转嫁至新销售；其余结转、补发、累计补发等一律不发了。'
+ws.merge_cells(start_row=rn, start_column=1, end_row=rn, end_column=3)
+ws.cell(row=rn, column=1, value=note_text).font = NOTE_FNT
+ws.cell(row=rn, column=1).alignment = Alignment(horizontal='left', vertical='center', wrap_text=True)
+ws.row_dimensions[rn].height = 36
+
+# ── Column widths ──
+ws.column_dimensions['A'].width = 16
+ws.column_dimensions['B'].width = 18
+ws.column_dimensions['C'].width = 16
+
+# ── Rename resigned sheets ──
+for old_name, new_name in [('二部-赵连飞','【已离职】二部-赵连飞'),('六部-伍仲文','【已离职】六部-伍仲文'),('六部-岑彦','【已离职】六部-岑彦')]:
+    if old_name in wb.sheetnames:
+        wb[old_name].title = new_name
+        wb.move_sheet(new_name, offset=len(wb.sheetnames)-1-wb.sheetnames.index(new_name))
+        print(f"  Sheet: {old_name} → {new_name}")
+
+# ── Copy to output ──
+import shutil
+out = '/Users/kkfluous/Downloads/20260713/里程任务考核_6月核算.xlsx'
+wb.save(out)
+print(f"\n✅ {out} 已重建")
+
+# ── Verify ──
+wb2 = openpyxl.load_workbook(out, data_only=True)
+ws2 = wb2['6月汇总']
+print(f"\n验证 ({ws2.max_row} rows):")
+for row in ws2.iter_rows(min_row=1, max_row=ws2.max_row):
+    vals = [str(c.value)[:50] if c.value is not None else '' for c in row]
+    line = ' | '.join(vals)
+    if line.replace(' | ', '').strip():
+        print(f"  {line}")
+wb2.close()
diff --git a/recalculate_july_assessment_mileage_daily.py b/recalculate_july_assessment_mileage_daily.py
new file mode 100644
index 0000000..d7d3f8a
--- /dev/null
+++ b/recalculate_july_assessment_mileage_daily.py
@@ -0,0 +1,203 @@
+from ast import parse, walk, Call, Constant, keyword
+from collections import defaultdict
+from copy import copy
+from datetime import datetime, date
+from pathlib import Path
+from zipfile import ZipFile
+from xml.etree import ElementTree as ET
+import re
+import shutil
+
+import openpyxl
+import pymysql
+from openpyxl.styles import Alignment, Font, PatternFill
+
+BASE = Path('/Users/kkfluous/Downloads')
+OUTPUT = BASE / '7月核算结果/7月考核车辆多源里程汇总_含车型.xlsx'
+MAIN_OUTPUT = BASE / '7月核算结果/7月考核车辆多源里程汇总.xlsx'
+BACKUP = BASE / '7月核算结果/7月考核车辆多源里程汇总_新日规则前.xlsx'
+ASSESSMENT = BASE / '7月核算结果/租赁任务考核_2026年7月.xlsx'
+TBOX_FILE = BASE / '车辆里程查询_20260701-20260731_1024辆.xlsx'
+PLATFORM_DIR = BASE / '广州新能源平台_纯氢纯电里程_2026年7月_按日'
+DETAILS = [BASE / '里程统计[天][2026-07-01至2026-07-31].xlsx', BASE / '里程统计[天][2026-07-01至2026-07-31] (1).xlsx']
+NS = '{http://schemas.openxmlformats.org/spreadsheetml/2006/main}'
+
+def norm(v): return re.sub(r'\s+', '', str(v or '')).upper()
+
+def db_config():
+    tree = parse((BASE / 'replace_gps_jun.py').read_text())
+    for node in walk(tree):
+        if isinstance(node, Call) and getattr(node.func, 'attr', None) == 'connect':
+            values = {x.arg: x.value.value for x in node.keywords if isinstance(x.value, Constant)}
+            if {'host', 'user', 'password', 'database'} <= values.keys(): return values
+    raise RuntimeError('未找到既有GPS数据库配置')
+
+def read_detail_gps():
+    out = defaultdict(dict)
+    for path in DETAILS:
+        with ZipFile(path) as zf:
+            root = ET.fromstring(zf.read('xl/worksheets/sheet1.xml'))
+            for row in root.findall(f'.//{NS}row'):
+                if int(row.get('r')) < 4: continue
+                values = {}
+                for cell in row.findall(f'{NS}c'):
+                    col = re.match(r'[A-Z]+', cell.get('r')).group()
+                    if cell.get('t') == 'inlineStr':
+                        values[col] = ''.join(t.text or '' for t in cell.findall(f'.//{NS}t'))
+                    else:
+                        node = cell.find(f'{NS}v'); values[col] = node.text if node is not None else None
+                p = norm(values.get('H')); ds = str(values.get('O') or '').strip()
+                if p and p != '合计' and re.fullmatch(r'2026-07-\d{2}', ds):
+                    out[p][datetime.strptime(ds, '%Y-%m-%d').date()] = float(values.get('P') or 0)
+    return out
+
+# 测试候选车辆以7月最终考核源表中黄色干预行为准。
+awb = openpyxl.load_workbook(ASSESSMENT, data_only=True)
+aws = awb['业务考核视图']
+candidates = set()
+for row in range(2, aws.max_row + 1):
+    fills = {aws.cell(row, c).fill.fgColor.rgb for c in range(1, aws.max_column + 1) if aws.cell(row, c).fill.fill_type == 'solid'}
+    if any(str(x).upper().endswith('FFFF00') for x in fills if x):
+        candidates.add(norm(aws.cell(row, 1).value))
+awb.close()
+
+# 每日TBOX。文件中的空白日按0处理，不会触发测试条件。
+twb = openpyxl.load_workbook(TBOX_FILE, read_only=True, data_only=True)
+tws = twb['里程查询']
+header = next(tws.iter_rows(min_row=6, max_row=6, values_only=True))
+day_cols = {i: value.date() for i, value in enumerate(header) if isinstance(value, datetime)}
+tbox = defaultdict(dict)
+for row in tws.iter_rows(min_row=7, values_only=True):
+    p = norm(row[0])
+    if p:
+        for i, day in day_cols.items(): tbox[p][day] = float(row[i] or 0)
+twb.close()
+
+# 每日GPS：两份日明细优先，数据库补充缺失日期。
+gps = read_detail_gps()
+cfg = db_config(); cfg.update(charset='utf8mb4', connect_timeout=8, read_timeout=30, write_timeout=30)
+conn = pymysql.connect(**cfg)
+try:
+    with conn.cursor() as cur:
+        placeholders = ','.join(['%s'] * len(candidates))
+        cur.execute(
+            f'SELECT plate_number,dates,total_mileage FROM ln_vehicle_g7_mileage WHERE dates >= %s AND dates <= %s AND plate_number IN ({placeholders})',
+            [date(2026, 7, 1), date(2026, 7, 31), *sorted(candidates)],
+        )
+        db_rows = cur.fetchall()
+finally: conn.close()
+for p, day, raw in db_rows:
+    if isinstance(day, datetime): day = day.date()
+    gps[norm(p)].setdefault(day, float(raw or 0) / 100000.0)
+
+# 广州平台每日纯氢里程。
+hydrogen = defaultdict(dict)
+for path in sorted(PLATFORM_DIR.glob('2026-07-??_纯氢纯电里程.xlsx')):
+    wb = openpyxl.load_workbook(path, read_only=True, data_only=True)
+    ws = wb.active
+    for row in ws.iter_rows(min_row=2, values_only=True):
+        p = norm(row[5]); match = re.search(r'2026-07-\d{2}', str(row[6] or ''))
+        if p and match:
+            day = datetime.strptime(match.group(), '%Y-%m-%d').date()
+            hydrogen[p][day] = hydrogen[p].get(day, 0) + float(row[8] or 0)
+    wb.close()
+
+if not BACKUP.exists(): shutil.copy2(OUTPUT, BACKUP)
+wb = openpyxl.load_workbook(OUTPUT)
+ws = wb.active
+header_row = next(r for r in range(1, 8) if ws.cell(r, 2).value == '车牌')
+headers = [ws.cell(header_row, c).value for c in range(1, ws.max_column + 1)]
+if '考核里程(km)' not in headers:
+    for merged in list(ws.merged_cells.ranges):
+        if merged.min_row == 1 and merged.max_row == 1:
+            ws.unmerge_cells(str(merged))
+    total_col = headers.index('广州平台总里程(km)') + 1
+    ws.insert_cols(total_col + 1, 1)
+    assessment_col = total_col + 1
+else:
+    assessment_col = headers.index('考核里程(km)') + 1
+ws.cell(header_row, assessment_col).value = '考核里程(km)'
+ws.cell(header_row, assessment_col)._style = copy(ws.cell(header_row, assessment_col - 1)._style)
+ws.column_dimensions[openpyxl.utils.get_column_letter(assessment_col)].width = 18
+
+headers = [ws.cell(header_row, c).value for c in range(1, ws.max_column + 1)]
+col = {v: i + 1 for i, v in enumerate(headers)}
+yes_records = 0; test_days_total = 0; excluded_hydrogen_total = 0.0; no_hydrogen = 0
+fallback_tbox = 0; fallback_gps = 0; fallback_missing = 0
+white = PatternFill('solid', fgColor='FFFFFF')
+platform_fill = PatternFill('solid', fgColor='DDEBF7')
+assessment_fill = PatternFill('solid', fgColor='FCE4D6')
+tbox_fill = PatternFill('solid', fgColor='E2F0D9')
+gps_fill = PatternFill('solid', fgColor='E4DFEC')
+yellow = PatternFill('solid', fgColor='FFF2CC')
+grey = PatternFill('solid', fgColor='F2F2F2')
+
+for row in range(header_row + 1, ws.max_row + 1):
+    p = norm(ws.cell(row, col['车牌']).value)
+    start = ws.cell(row, col['考核开始日期']).value.date()
+    end = ws.cell(row, col['考核结束日期']).value.date()
+    days = [start.fromordinal(start.toordinal() + n) for n in range((end - start).days + 1)]
+    test_days = []
+    if p in candidates:
+        for day in days:
+            if day in gps.get(p, {}) and tbox[p].get(day, 0) - gps[p][day] > 50:
+                test_days.append(day)
+    has_test = bool(test_days)
+    ws.cell(row, col['是否存在测试']).value = '是' if has_test else '否'
+    if has_test: yes_records += 1
+    test_days_total += len(test_days)
+    daily_h = hydrogen.get(p, {})
+    present = [day for day in days if day in daily_h]
+    if not present:
+        no_hydrogen += 1
+        fallback_name = 'GPS里程(km)' if p in candidates else 'TBOX仪表里程(km)'
+        fallback_value = ws.cell(row, col[fallback_name]).value
+        if isinstance(fallback_value, (int, float)):
+            ws.cell(row, assessment_col).value = round(float(fallback_value), 2)
+            ws.cell(row, assessment_col).number_format = '#,##0.0'
+            if p in candidates:
+                fallback_gps += 1
+            else:
+                fallback_tbox += 1
+        else:
+            ws.cell(row, assessment_col).value = '无来源'
+            fallback_missing += 1
+    else:
+        excluded_hydrogen_total += sum(daily_h.get(day, 0) for day in test_days)
+        ws.cell(row, assessment_col).value = round(sum(daily_h.get(day, 0) for day in days if day not in test_days), 2)
+        ws.cell(row, assessment_col).number_format = '#,##0.0'
+
+    # 重建行配色：新规则判定存在测试时整行黄色，否则恢复各数据组颜色。
+    if has_test:
+        for c in range(1, ws.max_column + 1): ws.cell(row, c).fill = copy(yellow)
+    else:
+        for c in range(1, ws.max_column + 1): ws.cell(row, c).fill = copy(white)
+        for name in ['纯氢里程(km)', '纯电里程(km)', '广州平台总里程(km)']:
+            ws.cell(row, col[name]).fill = copy(platform_fill)
+        ws.cell(row, assessment_col).fill = copy(assessment_fill)
+        ws.cell(row, col['TBOX仪表里程(km)']).fill = copy(tbox_fill)
+        ws.cell(row, col['GPS里程(km)']).fill = copy(gps_fill)
+        for name in ['纯氢里程(km)', '纯电里程(km)', '广州平台总里程(km)', '考核里程(km)', 'TBOX仪表里程(km)', 'GPS里程(km)']:
+            cell = ws.cell(row, col[name])
+            if cell.value == '无来源':
+                cell.fill = copy(grey); cell.font = Font(color='7F7F7F', italic=True)
+
+for merged in list(ws.merged_cells.ranges):
+    if merged.min_row == 1 and merged.max_row == 1:
+        ws.unmerge_cells(str(merged))
+ws.merge_cells(start_row=1, start_column=1, end_row=1, end_column=ws.max_column)
+ws['A1'] = (
+    '1. 考核里程按纯氢里程计算：普通车辆取考核区间纯氢里程；测试候选车辆逐日比较，'
+    '当日TBOX里程－当日GPS里程＞50km时判为测试日，并剔除该日纯氢里程；'
+    '无纯氢里程的4.5T普货车辆，测试候选车辆取GPS里程，其他车辆取TBOX仪表里程\n'
+    '2. 每日TBOX来源：车辆里程查询_20260701-20260731_1024辆.xlsx；空白日按0处理。每日GPS优先两份7月日明细，数据库补充\n'
+    '3. 4.5T普货的TBOX或GPS备选来源也缺失时，考核里程标记“无来源”\n'
+    '4. 纯电里程及广州平台总里程仅作对照，不参与考核里程'
+)
+ws['A1'].alignment = Alignment(horizontal='left', vertical='center', wrap_text=True)
+ws.auto_filter.ref = f'A{header_row}:{openpyxl.utils.get_column_letter(ws.max_column)}{ws.max_row}'
+ws.row_dimensions[1].height = 58
+wb.save(OUTPUT)
+shutil.copy2(OUTPUT, MAIN_OUTPUT)
+
+print({'candidate_plates': len(candidates), 'test_records_new_rule': yes_records, 'test_days': test_days_total, 'excluded_hydrogen_km': round(excluded_hydrogen_total, 2), 'pure_hydrogen_missing': no_hydrogen, 'fallback_tbox_records': fallback_tbox, 'fallback_gps_records': fallback_gps, 'assessment_mileage_still_missing': fallback_missing, 'columns': ws.max_column})
diff --git a/regenerate_intervention.py b/regenerate_intervention.py
new file mode 100644
index 0000000..6096e87
--- /dev/null
+++ b/regenerate_intervention.py
@@ -0,0 +1,270 @@
+#!/usr/bin/env python3
+"""用新GPS数据重新生成6月干预情况说明 — 测试里程/实际里程/翻转分析全部更新"""
+import openpyxl, os
+from openpyxl.styles import Font, Alignment, Border, Side, PatternFill
+from openpyxl.utils import get_column_letter
+from collections import defaultdict
+
+os.chdir('/Users/kkfluous/Downloads')
+
+# ── Styles ──
+hdr_blue_font = Font(name='宋体', size=11, bold=True, color='FFFFFF')
+hdr_blue_fill = PatternFill(start_color='4472C4', end_color='4472C4', fill_type='solid')
+hdr_green_font = Font(name='宋体', size=11, bold=True, color='FFFFFF')
+hdr_green_fill = PatternFill(start_color='548235', end_color='548235', fill_type='solid')
+hdr_align = Alignment(horizontal='center', vertical='center', wrap_text=True)
+data_font = Font(name='宋体', size=10)
+data_align = Alignment(horizontal='center', vertical='center')
+title_font = Font(name='宋体', size=14, bold=True)
+subtitle_font = Font(name='宋体', size=10, bold=True)
+yellow_fill = PatternFill(start_color='FFFF00', end_color='FFFF00', fill_type='solid')
+light_yellow_fill = PatternFill(start_color='FFF2CC', end_color='FFF2CC', fill_type='solid')
+thin_border = Border(left=Side('thin'), right=Side('thin'), top=Side('thin'), bottom=Side('thin'))
+
+def set_cell(ws, row, col, value, font=None, fill=None, align=None, border=None, num_fmt=None):
+    cell = ws.cell(row=row, column=col, value=value)
+    if font: cell.font = font
+    if fill: cell.fill = fill
+    if align: cell.alignment = align
+    if border: cell.border = border
+    if num_fmt: cell.number_format = num_fmt
+    return cell
+
+# ── Step 1: Extract test plates from OLD intervention report ──
+print("提取测试车辆清单...")
+old_interv = '租赁任务考核_2026年6月_干预情况说明.xlsx'
+wb_old = openpyxl.load_workbook(old_interv, data_only=True)
+ws_old = wb_old['涉及测试的业务考核车辆']
+h_old = [c.value for c in next(ws_old.iter_rows(min_row=1, max_row=1))]
+plate_col_old = h_old.index('车牌号')
+test_plates = set()
+for row in ws_old.iter_rows(min_row=2, values_only=True):
+    p = row[plate_col_old]
+    if p: test_plates.add(str(p).strip())
+wb_old.close()
+print(f"  测试车辆: {len(test_plates)} 辆")
+
+# ── Step 2: Read NEW source data ──
+print("读取新数据源...")
+src_fp = '20260713_1500/租赁任务考核_2026年6月.xlsx'
+src = openpyxl.load_workbook(src_fp, data_only=True)
+ws_src = src['业务考核视图']
+h_src = [c.value for c in next(ws_src.iter_rows(min_row=1, max_row=1))]
+all_recs = []
+for row in ws_src.iter_rows(min_row=2, values_only=True):
+    r = dict(zip(h_src, row))
+    if r['车牌号']:
+        all_recs.append(r)
+src.close()
+
+# ── Step 3: Read BACKUP for original comparison ──
+print("读取备份(原始仪表盘)...")
+bak = openpyxl.load_workbook('租赁任务考核_2026年6月_备份.xlsx', data_only=True)
+ws_bak = bak['业务考核视图']
+h_bak = [c.value for c in next(ws_bak.iter_rows(min_row=1, max_row=1))]
+bak_recs = {}
+for row in ws_bak.iter_rows(min_row=2, values_only=True):
+    r = dict(zip(h_bak, row))
+    if r['车牌号']:
+        bak_recs[str(r['车牌号']).strip()] = r
+bak.close()
+
+# ── Step 4: Analyze ──
+print("分析干预数据...")
+intervention_recs = [r for r in all_recs if r['车牌号'] in test_plates]
+intervention_plates = set(r['车牌号'] for r in intervention_recs)
+
+# Stats
+total_vehicles = len(set(r['车牌号'] for r in all_recs))
+total_records = len(all_recs)
+interv_vehicles = len(intervention_plates)
+interv_records = len(intervention_recs)
+
+total_actual = sum(float(r['实际行驶里程(km)'] or 0) for r in all_recs)
+total_tbox = sum(float(r['原TBOX里程(km)'] or 0) for r in all_recs)
+total_test = sum(float(r['测试里程(km)'] or 0) for r in all_recs)
+interv_actual = sum(float(r['实际行驶里程(km)'] or 0) for r in intervention_recs)
+interv_tbox = sum(float(r['原TBOX里程(km)'] or 0) for r in intervention_recs)
+interv_test = sum(float(r['测试里程(km)'] or 0) for r in intervention_recs)
+
+# Flip analysis: old达→new未达标
+flip_recs = []
+for r in intervention_recs:
+    plate = r['车牌号']
+    old_rec = bak_recs.get(plate)
+    if old_rec:
+        old_q = str(old_rec.get('是否达标', '')).strip() == '达标'
+        new_q = str(r.get('是否达标', '')).strip() == '达标'
+        if old_q and not new_q:
+            flip_recs.append(r)
+
+# Group flip by target
+flip_by_target = defaultdict(lambda: defaultdict(lambda: defaultdict(int)))
+for r in flip_recs:
+    target = r.get('考核目标', '未知')
+    dept = r.get('部门名称', '')
+    sales = r.get('销售经理', '')
+    flip_by_target[target][dept][sales] += 1
+
+print(f"  全量: {total_vehicles}辆/{total_records}条")
+print(f"  干预: {interv_vehicles}辆/{interv_records}条")
+print(f"  达→未达标: {len(flip_recs)}条")
+print(f"  TBOX总计: {total_tbox:.0f}, GPS总计: {total_actual:.0f}, 测试总计: {total_test:.0f}")
+
+# ── Step 5: Build workbook ──
+wb = openpyxl.Workbook()
+
+# ===== Sheet 1: 测试与考核情况说明 =====
+ws1 = wb.active; ws1.title = '测试与考核情况说明'
+for col, w in enumerate([4, 30, 18, 18, 18, 18, 18], 1):
+    ws1.column_dimensions[get_column_letter(col)].width = w
+
+rn = 1
+set_cell(ws1, rn, 1, '2026年6月 业务考核 — 测试与考核情况说明', title_font)
+ws1.merge_cells(start_row=rn, start_column=1, end_row=rn, end_column=5); rn += 1
+
+set_cell(ws1, rn, 1, '6月业务考核车辆中，部分车辆当月存在测试行为（仪表盘里程 > GPS 实际行驶里程），其差额作为测试里程剥离，不参与考核。GPS数据来源：G7/东方北斗/信达GPS/广安北斗 四家供应商6月日里程汇总文件。',
+         Font(name='宋体', size=10), light_yellow_fill)
+ws1.merge_cells(start_row=rn, start_column=1, end_row=rn, end_column=5); rn += 2
+
+# Section 1: Stats
+set_cell(ws1, rn, 1, '一、6月业务考核与干预情况', subtitle_font); rn += 1
+for ci, h in enumerate(['项目', '6月业务考核车辆', '涉及测试的车辆', '占比'], 1):
+    set_cell(ws1, rn, ci, h, hdr_blue_font, hdr_blue_fill, hdr_align, thin_border)
+rn += 1
+
+stats = [
+    ('车辆数(辆)', total_vehicles, interv_vehicles),
+    ('产生考核记录数(条)', total_records, interv_records),
+    ('仪表盘里程(km)', round(total_tbox), round(interv_tbox)),
+    ('业务考核里程 GPS(km)', round(total_actual), round(interv_actual)),
+    ('测试里程(km)', round(total_test), round(interv_test)),
+]
+for label, all_v, interv_v in stats:
+    ratio = interv_v / all_v if all_v else 0
+    set_cell(ws1, rn, 1, label, data_font, None, data_align, thin_border)
+    set_cell(ws1, rn, 2, all_v, data_font, None, data_align, thin_border)
+    set_cell(ws1, rn, 3, interv_v, data_font, None, data_align, thin_border)
+    set_cell(ws1, rn, 4, ratio, data_font, None, data_align, thin_border, '0.00%')
+    rn += 1
+
+rn += 2
+
+# Section 2: Flip details
+set_cell(ws1, rn, 1, '二、干预车辆考核达标情况说明', subtitle_font); rn += 1
+flip_total = len(flip_recs)
+summary_text = f'共 {interv_records} 条考核记录涉及里程干预；替换为GPS里程后，共 {flip_total} 条考核车辆记录由"达标"转判为"未达标"。按车型分类如下：'
+set_cell(ws1, rn, 1, summary_text, Font(name='宋体', size=10))
+ws1.merge_cells(start_row=rn, start_column=1, end_row=rn, end_column=5); rn += 2
+
+target_order = ['羚牛136辆4.5T冷链车', '羚牛100辆18T', '交投40辆4.5T普货', '交投190辆4.5T冷链车', '恒运50辆4.5T普货']
+for target in target_order:
+    if target not in flip_by_target: continue
+    dept_sales = flip_by_target[target]
+    total_cnt = sum(sum(sales.values()) for sales in dept_sales.values())
+    set_cell(ws1, rn, 1, f'▎{target}（{total_cnt} 辆）', subtitle_font)
+    ws1.merge_cells(start_row=rn, start_column=1, end_row=rn, end_column=5); rn += 1
+
+    set_cell(ws1, rn, 1, '部门', hdr_green_font, hdr_green_fill, hdr_align, thin_border)
+    set_cell(ws1, rn, 2, '销售经理', hdr_green_font, hdr_green_fill, hdr_align, thin_border)
+    set_cell(ws1, rn, 3, '数量', hdr_green_font, hdr_green_fill, hdr_align, thin_border)
+    set_cell(ws1, rn, 4, '部门', hdr_green_font, hdr_green_fill, hdr_align, thin_border)
+    set_cell(ws1, rn, 5, '销售经理', hdr_green_font, hdr_green_fill, hdr_align, thin_border)
+    set_cell(ws1, rn, 6, '数量', hdr_green_font, hdr_green_fill, hdr_align, thin_border)
+    rn += 1
+
+    items = []
+    for dept in sorted(dept_sales.keys()):
+        for sales in sorted(dept_sales[dept].keys()):
+            items.append((dept, sales, dept_sales[dept][sales]))
+
+    mid = (len(items) + 1) // 2
+    max_rows = max(mid, len(items) - mid)
+    for i in range(max_rows):
+        if i < mid:
+            d1, s1, c1 = items[i]
+            set_cell(ws1, rn, 1, d1, data_font, None, data_align, thin_border)
+            set_cell(ws1, rn, 2, s1, data_font, None, data_align, thin_border)
+            set_cell(ws1, rn, 3, c1, data_font, None, data_align, thin_border)
+        else:
+            for c in [1,2,3]: set_cell(ws1, rn, c, '', data_font, None, data_align, thin_border)
+        if i + mid < len(items):
+            d2, s2, c2 = items[i + mid]
+            set_cell(ws1, rn, 4, d2, data_font, None, data_align, thin_border)
+            set_cell(ws1, rn, 5, s2, data_font, None, data_align, thin_border)
+            set_cell(ws1, rn, 6, c2, data_font, None, data_align, thin_border)
+        else:
+            for c in [4,5,6]: set_cell(ws1, rn, c, '', data_font, None, data_align, thin_border)
+        rn += 1
+    rn += 1
+
+rn += 1
+
+# Section 3: Column source explanation
+set_cell(ws1, rn, 1, '三、里程三列数据来源说明', subtitle_font); rn += 1
+set_cell(ws1, rn, 1, '列', hdr_blue_font, hdr_blue_fill, hdr_align, thin_border)
+set_cell(ws1, rn, 2, '字段名', hdr_blue_font, hdr_blue_fill, hdr_align, thin_border)
+set_cell(ws1, rn, 3, '数据来源', hdr_blue_font, hdr_blue_fill, hdr_align, thin_border)
+set_cell(ws1, rn, 4, '说明', hdr_blue_font, hdr_blue_fill, hdr_align, thin_border)
+rn += 1
+
+col_descs = [
+    ('O', '原TBOX里程(km)', '车辆 T-Box / 仪表盘读数', '当月仪表总里程（业务系统原始数据），含运营+测试+调试里程。仅干预车辆填充，非干预车辆为空'),
+    ('P', '实际行驶里程(km)', 'GPS 轨迹还原', '基于 G7/东方北斗/信达GPS/广安北斗 四家供应商6月GPS日里程汇总文件，按考核期逐日汇总计算的运营里程，本月里程考核以此列为准'),
+    ('Q', '测试里程(km)', 'O − P（原TBOX − GPS，差额）', '仪表盘跑了但 GPS 未记录到的部分，判定为厂内测试/调试/路试等非运营里程，不计入考核。仅干预车辆填充，非干预车辆为0。若差额为负则取0'),
+]
+for col_letter, name, source, desc in col_descs:
+    set_cell(ws1, rn, 1, col_letter, data_font, None, data_align, thin_border)
+    set_cell(ws1, rn, 2, name, data_font, None, Alignment(horizontal='left', vertical='center'), thin_border)
+    set_cell(ws1, rn, 3, source, data_font, None, data_align, thin_border)
+    set_cell(ws1, rn, 4, desc, data_font, None, Alignment(horizontal='left', vertical='center', wrap_text=True), thin_border)
+    ws1.row_dimensions[rn].height = 40
+    rn += 1
+
+# ===== Sheet 2: 业务考核视图 (full data,干预标黄) =====
+ws2 = wb.create_sheet('业务考核视图')
+target_order_full = ['车牌号','考核年份','考核月份','部门名称','销售经理','客户名称','合同编号','项目名称',
+                     '考核目标ID','考核目标','考核开始日期','考核结束日期','考核天数','应考核里程(km)',
+                     '原TBOX里程(km)','实际行驶里程(km)','测试里程(km)','完成率(%)','是否达标','考核状态',
+                     '交车日期','预计还车日期','实际还车日期']
+for ci, name in enumerate(target_order_full, 1):
+    set_cell(ws2, 1, ci, name, hdr_blue_font, hdr_blue_fill, hdr_align, thin_border)
+for ri, r in enumerate(all_recs, 2):
+    is_interv = r['车牌号'] in intervention_plates
+    for ci, name in enumerate(target_order_full, 1):
+        set_cell(ws2, ri, ci, r.get(name), data_font, yellow_fill if is_interv else None, data_align, thin_border)
+for col, w in enumerate([12,8,8,12,10,26,18,22,10,22,13,12,10,14,16,14,14,12,10,8,12,12,12], 1):
+    ws2.column_dimensions[get_column_letter(col)].width = w
+
+# ===== Sheet 3: 6月业务考核车辆总览 =====
+ws3 = wb.create_sheet('6月业务考核车辆总览')
+for ci, name in enumerate(target_order_full, 1):
+    set_cell(ws3, 1, ci, name, hdr_blue_font, hdr_blue_fill, hdr_align, thin_border)
+for ri, r in enumerate(all_recs, 2):
+    for ci, name in enumerate(target_order_full, 1):
+        set_cell(ws3, ri, ci, r.get(name), data_font, None, data_align, thin_border)
+for col, w in enumerate([12,8,8,12,10,26,18,22,10,22,13,12,10,14,16,14,14,12,10,8,12,12,12], 1):
+    ws3.column_dimensions[get_column_letter(col)].width = w
+
+# ===== Sheet 4: 涉及测试的业务考核车辆 =====
+ws4 = wb.create_sheet('涉及测试的业务考核车辆')
+for ci, name in enumerate(target_order_full, 1):
+    set_cell(ws4, 1, ci, name, hdr_blue_font, hdr_blue_fill, hdr_align, thin_border)
+for ri, r in enumerate(intervention_recs, 2):
+    for ci, name in enumerate(target_order_full, 1):
+        set_cell(ws4, ri, ci, r.get(name), data_font, yellow_fill, data_align, thin_border)
+for col, w in enumerate([12,8,8,12,10,26,18,22,10,22,13,12,10,14,16,14,14,12,10,8,12,12,12], 1):
+    ws4.column_dimensions[get_column_letter(col)].width = w
+
+# ── Save ──
+out = '20260713_1500/租赁任务考核_2026年6月_干预情况说明.xlsx'
+wb.save(out)
+
+print(f'\n✅ {out}')
+print(f'\n概况:')
+print(f'  全量: {total_vehicles}辆, {total_records}条')
+print(f'  干预: {interv_vehicles}辆, {interv_records}条')
+print(f'  达→未达标: {flip_total}条')
+print(f'  仪表盘里程: {total_tbox:.0f} (干预{interv_tbox:.0f})')
+print(f'  GPS里程: {total_actual:.0f} (干预{interv_actual:.0f})')
+print(f'  测试里程: {total_test:.0f} (干预{interv_test:.0f})')
diff --git a/regenerate_intervention_jul.py b/regenerate_intervention_jul.py
new file mode 100644
index 0000000..87894f8
--- /dev/null
+++ b/regenerate_intervention_jul.py
@@ -0,0 +1,269 @@
+#!/usr/bin/env python3
+"""用新GPS数据重新生成7月干预情况说明 — 测试里程/实际里程/翻转分析全部更新"""
+import openpyxl, os
+from openpyxl.styles import Font, Alignment, Border, Side, PatternFill
+from openpyxl.utils import get_column_letter
+from collections import defaultdict
+
+os.chdir('/Users/kkfluous/Downloads')
+
+# ── Styles ──
+hdr_blue_font = Font(name='宋体', size=11, bold=True, color='FFFFFF')
+hdr_blue_fill = PatternFill(start_color='4472C4', end_color='4472C4', fill_type='solid')
+hdr_green_font = Font(name='宋体', size=11, bold=True, color='FFFFFF')
+hdr_green_fill = PatternFill(start_color='548235', end_color='548235', fill_type='solid')
+hdr_align = Alignment(horizontal='center', vertical='center', wrap_text=True)
+data_font = Font(name='宋体', size=10)
+data_align = Alignment(horizontal='center', vertical='center')
+title_font = Font(name='宋体', size=14, bold=True)
+subtitle_font = Font(name='宋体', size=10, bold=True)
+yellow_fill = PatternFill(start_color='FFFF00', end_color='FFFF00', fill_type='solid')
+light_yellow_fill = PatternFill(start_color='FFF2CC', end_color='FFF2CC', fill_type='solid')
+thin_border = Border(left=Side('thin'), right=Side('thin'), top=Side('thin'), bottom=Side('thin'))
+
+def set_cell(ws, row, col, value, font=None, fill=None, align=None, border=None, num_fmt=None):
+    cell = ws.cell(row=row, column=col, value=value)
+    if font: cell.font = font
+    if fill: cell.fill = fill
+    if align: cell.alignment = align
+    if border: cell.border = border
+    if num_fmt: cell.number_format = num_fmt
+    return cell
+
+# ── Step 1: Extract test plates from July test list ──
+print("提取测试车辆清单...")
+test_file = '/Users/kkfluous/Library/Containers/com.tencent.xinWeChat/Data/Documents/xwechat_files/wxid_1704407055112_3af8/temp/RWTemp/2026-08/4340099467fbf08b4c6583c49adba4ee/7月测试车辆明细.xlsx'
+wb_old = openpyxl.load_workbook(test_file, data_only=True)
+ws_old = wb_old.active
+test_plates = set()
+for row in ws_old.iter_rows(min_row=2, values_only=True):
+    p = row[0]
+    if p: test_plates.add(str(p).strip())
+test_plates.discard('粤AG37316')
+wb_old.close()
+print(f"  测试车辆: {len(test_plates)} 辆")
+
+# ── Step 2: Read NEW source data ──
+print("读取新数据源...")
+src_fp = '7月核算结果/租赁任务考核_2026年7月.xlsx'
+src = openpyxl.load_workbook(src_fp, data_only=True)
+ws_src = src['业务考核视图']
+h_src = [c.value for c in next(ws_src.iter_rows(min_row=1, max_row=1))]
+all_recs = []
+for row in ws_src.iter_rows(min_row=2, values_only=True):
+    r = dict(zip(h_src, row))
+    if r['车牌号']:
+        all_recs.append(r)
+src.close()
+
+# ── Step 3: Read BACKUP for original comparison ──
+print("读取备份(原始仪表盘)...")
+bak = openpyxl.load_workbook('7月核算结果/租赁任务考核_2026年7月_原版备份.xlsx', data_only=True)
+ws_bak = bak['业务考核视图']
+h_bak = [c.value for c in next(ws_bak.iter_rows(min_row=1, max_row=1))]
+bak_recs = {}
+for row in ws_bak.iter_rows(min_row=2, values_only=True):
+    r = dict(zip(h_bak, row))
+    if r['车牌号']:
+        bak_recs[str(r['车牌号']).strip()] = r
+bak.close()
+
+# ── Step 4: Analyze ──
+print("分析干预数据...")
+intervention_recs = [r for r in all_recs if r['车牌号'] in test_plates]
+intervention_plates = set(r['车牌号'] for r in intervention_recs)
+
+# Stats
+total_vehicles = len(set(r['车牌号'] for r in all_recs))
+total_records = len(all_recs)
+interv_vehicles = len(intervention_plates)
+interv_records = len(intervention_recs)
+
+total_actual = sum(float(r['实际行驶里程(km)'] or 0) for r in all_recs)
+total_tbox = sum(float(r['原TBOX里程(km)'] or 0) for r in all_recs)
+total_test = sum(float(r['测试里程(km)'] or 0) for r in all_recs)
+interv_actual = sum(float(r['实际行驶里程(km)'] or 0) for r in intervention_recs)
+interv_tbox = sum(float(r['原TBOX里程(km)'] or 0) for r in intervention_recs)
+interv_test = sum(float(r['测试里程(km)'] or 0) for r in intervention_recs)
+
+# Flip analysis: old达→new未达标
+flip_recs = []
+for r in intervention_recs:
+    plate = r['车牌号']
+    old_rec = bak_recs.get(plate)
+    if old_rec:
+        old_q = str(old_rec.get('是否达标', '')).strip() == '达标'
+        new_q = str(r.get('是否达标', '')).strip() == '达标'
+        if old_q and not new_q:
+            flip_recs.append(r)
+
+# Group flip by target
+flip_by_target = defaultdict(lambda: defaultdict(lambda: defaultdict(int)))
+for r in flip_recs:
+    target = r.get('考核目标', '未知')
+    dept = r.get('部门名称', '')
+    sales = r.get('销售经理', '')
+    flip_by_target[target][dept][sales] += 1
+
+print(f"  全量: {total_vehicles}辆/{total_records}条")
+print(f"  干预: {interv_vehicles}辆/{interv_records}条")
+print(f"  达→未达标: {len(flip_recs)}条")
+print(f"  TBOX总计: {total_tbox:.0f}, GPS总计: {total_actual:.0f}, 测试总计: {total_test:.0f}")
+
+# ── Step 5: Build workbook ──
+wb = openpyxl.Workbook()
+
+# ===== Sheet 1: 测试与考核情况说明 =====
+ws1 = wb.active; ws1.title = '测试与考核情况说明'
+for col, w in enumerate([4, 30, 18, 18, 18, 18, 18], 1):
+    ws1.column_dimensions[get_column_letter(col)].width = w
+
+rn = 1
+set_cell(ws1, rn, 1, '2026年7月 业务考核 — 测试与考核情况说明', title_font)
+ws1.merge_cells(start_row=rn, start_column=1, end_row=rn, end_column=5); rn += 1
+
+set_cell(ws1, rn, 1, '7月业务考核车辆中，测试车辆统一以指定GPS日里程按考核期汇总替换TBOX里程，差额作为测试里程剥离且最低取0。粤AG37316因GPS缺失，经业务确认暂按0核算。',
+         Font(name='宋体', size=10), light_yellow_fill)
+ws1.merge_cells(start_row=rn, start_column=1, end_row=rn, end_column=5); rn += 2
+
+# Section 1: Stats
+set_cell(ws1, rn, 1, '一、7月业务考核与干预情况', subtitle_font); rn += 1
+for ci, h in enumerate(['项目', '7月业务考核车辆', '涉及测试的车辆', '占比'], 1):
+    set_cell(ws1, rn, ci, h, hdr_blue_font, hdr_blue_fill, hdr_align, thin_border)
+rn += 1
+
+stats = [
+    ('车辆数(辆)', total_vehicles, interv_vehicles),
+    ('产生考核记录数(条)', total_records, interv_records),
+    ('仪表盘里程(km)', round(total_tbox), round(interv_tbox)),
+    ('业务考核里程 GPS(km)', round(total_actual), round(interv_actual)),
+    ('测试里程(km)', round(total_test), round(interv_test)),
+]
+for label, all_v, interv_v in stats:
+    ratio = interv_v / all_v if all_v else 0
+    set_cell(ws1, rn, 1, label, data_font, None, data_align, thin_border)
+    set_cell(ws1, rn, 2, all_v, data_font, None, data_align, thin_border)
+    set_cell(ws1, rn, 3, interv_v, data_font, None, data_align, thin_border)
+    set_cell(ws1, rn, 4, ratio, data_font, None, data_align, thin_border, '0.00%')
+    rn += 1
+
+rn += 2
+
+# Section 2: Flip details
+set_cell(ws1, rn, 1, '二、干预车辆考核达标情况说明', subtitle_font); rn += 1
+flip_total = len(flip_recs)
+summary_text = f'共 {interv_records} 条考核记录涉及里程干预；替换为GPS里程后，共 {flip_total} 条考核车辆记录由"达标"转判为"未达标"。按车型分类如下：'
+set_cell(ws1, rn, 1, summary_text, Font(name='宋体', size=10))
+ws1.merge_cells(start_row=rn, start_column=1, end_row=rn, end_column=5); rn += 2
+
+target_order = ['羚牛136辆4.5T冷链车', '羚牛100辆18T', '交投40辆4.5T普货', '交投190辆4.5T冷链车', '恒运50辆4.5T普货']
+for target in target_order:
+    if target not in flip_by_target: continue
+    dept_sales = flip_by_target[target]
+    total_cnt = sum(sum(sales.values()) for sales in dept_sales.values())
+    set_cell(ws1, rn, 1, f'▎{target}（{total_cnt} 辆）', subtitle_font)
+    ws1.merge_cells(start_row=rn, start_column=1, end_row=rn, end_column=5); rn += 1
+
+    set_cell(ws1, rn, 1, '部门', hdr_green_font, hdr_green_fill, hdr_align, thin_border)
+    set_cell(ws1, rn, 2, '销售经理', hdr_green_font, hdr_green_fill, hdr_align, thin_border)
+    set_cell(ws1, rn, 3, '数量', hdr_green_font, hdr_green_fill, hdr_align, thin_border)
+    set_cell(ws1, rn, 4, '部门', hdr_green_font, hdr_green_fill, hdr_align, thin_border)
+    set_cell(ws1, rn, 5, '销售经理', hdr_green_font, hdr_green_fill, hdr_align, thin_border)
+    set_cell(ws1, rn, 6, '数量', hdr_green_font, hdr_green_fill, hdr_align, thin_border)
+    rn += 1
+
+    items = []
+    for dept in sorted(dept_sales.keys()):
+        for sales in sorted(dept_sales[dept].keys()):
+            items.append((dept, sales, dept_sales[dept][sales]))
+
+    mid = (len(items) + 1) // 2
+    max_rows = max(mid, len(items) - mid)
+    for i in range(max_rows):
+        if i < mid:
+            d1, s1, c1 = items[i]
+            set_cell(ws1, rn, 1, d1, data_font, None, data_align, thin_border)
+            set_cell(ws1, rn, 2, s1, data_font, None, data_align, thin_border)
+            set_cell(ws1, rn, 3, c1, data_font, None, data_align, thin_border)
+        else:
+            for c in [1,2,3]: set_cell(ws1, rn, c, '', data_font, None, data_align, thin_border)
+        if i + mid < len(items):
+            d2, s2, c2 = items[i + mid]
+            set_cell(ws1, rn, 4, d2, data_font, None, data_align, thin_border)
+            set_cell(ws1, rn, 5, s2, data_font, None, data_align, thin_border)
+            set_cell(ws1, rn, 6, c2, data_font, None, data_align, thin_border)
+        else:
+            for c in [4,5,6]: set_cell(ws1, rn, c, '', data_font, None, data_align, thin_border)
+        rn += 1
+    rn += 1
+
+rn += 1
+
+# Section 3: Column source explanation
+set_cell(ws1, rn, 1, '三、里程三列数据来源说明', subtitle_font); rn += 1
+set_cell(ws1, rn, 1, '列', hdr_blue_font, hdr_blue_fill, hdr_align, thin_border)
+set_cell(ws1, rn, 2, '字段名', hdr_blue_font, hdr_blue_fill, hdr_align, thin_border)
+set_cell(ws1, rn, 3, '数据来源', hdr_blue_font, hdr_blue_fill, hdr_align, thin_border)
+set_cell(ws1, rn, 4, '说明', hdr_blue_font, hdr_blue_fill, hdr_align, thin_border)
+rn += 1
+
+col_descs = [
+    ('O', '原TBOX里程(km)', '车辆 T-Box / 仪表盘读数', '当月仪表总里程（业务系统原始数据），含运营+测试+调试里程。仅干预车辆填充，非干预车辆为空'),
+    ('P', '实际行驶里程(km)', '指定GPS日里程文件', '优先使用智能管车车辆里程日报；缺失车辆由两份里程统计日明细补充，并按考核期逐日汇总。粤AG37316经确认暂按0核算'),
+    ('Q', '测试里程(km)', 'O − P（原TBOX − GPS，差额）', '仪表盘跑了但 GPS 未记录到的部分，判定为厂内测试/调试/路试等非运营里程，不计入考核。仅干预车辆填充，非干预车辆为0。若差额为负则取0'),
+]
+for col_letter, name, source, desc in col_descs:
+    set_cell(ws1, rn, 1, col_letter, data_font, None, data_align, thin_border)
+    set_cell(ws1, rn, 2, name, data_font, None, Alignment(horizontal='left', vertical='center'), thin_border)
+    set_cell(ws1, rn, 3, source, data_font, None, data_align, thin_border)
+    set_cell(ws1, rn, 4, desc, data_font, None, Alignment(horizontal='left', vertical='center', wrap_text=True), thin_border)
+    ws1.row_dimensions[rn].height = 40
+    rn += 1
+
+# ===== Sheet 2: 业务考核视图 (full data,干预标黄) =====
+ws2 = wb.create_sheet('业务考核视图')
+target_order_full = ['车牌号','考核年份','考核月份','部门名称','销售经理','客户名称','合同编号','项目名称',
+                     '考核目标ID','考核目标','考核开始日期','考核结束日期','考核天数','应考核里程(km)',
+                     '原TBOX里程(km)','实际行驶里程(km)','测试里程(km)','完成率(%)','是否达标','考核状态',
+                     '交车日期','预计还车日期','实际还车日期']
+for ci, name in enumerate(target_order_full, 1):
+    set_cell(ws2, 1, ci, name, hdr_blue_font, hdr_blue_fill, hdr_align, thin_border)
+for ri, r in enumerate(all_recs, 2):
+    is_interv = r['车牌号'] in intervention_plates
+    for ci, name in enumerate(target_order_full, 1):
+        set_cell(ws2, ri, ci, r.get(name), data_font, yellow_fill if is_interv else None, data_align, thin_border)
+for col, w in enumerate([12,8,8,12,10,26,18,22,10,22,13,12,10,14,16,14,14,12,10,8,12,12,12], 1):
+    ws2.column_dimensions[get_column_letter(col)].width = w
+
+# ===== Sheet 3: 7月业务考核车辆总览 =====
+ws3 = wb.create_sheet('7月业务考核车辆总览')
+for ci, name in enumerate(target_order_full, 1):
+    set_cell(ws3, 1, ci, name, hdr_blue_font, hdr_blue_fill, hdr_align, thin_border)
+for ri, r in enumerate(all_recs, 2):
+    for ci, name in enumerate(target_order_full, 1):
+        set_cell(ws3, ri, ci, r.get(name), data_font, None, data_align, thin_border)
+for col, w in enumerate([12,8,8,12,10,26,18,22,10,22,13,12,10,14,16,14,14,12,10,8,12,12,12], 1):
+    ws3.column_dimensions[get_column_letter(col)].width = w
+
+# ===== Sheet 4: 涉及测试的业务考核车辆 =====
+ws4 = wb.create_sheet('涉及测试的业务考核车辆')
+for ci, name in enumerate(target_order_full, 1):
+    set_cell(ws4, 1, ci, name, hdr_blue_font, hdr_blue_fill, hdr_align, thin_border)
+for ri, r in enumerate(intervention_recs, 2):
+    for ci, name in enumerate(target_order_full, 1):
+        set_cell(ws4, ri, ci, r.get(name), data_font, yellow_fill, data_align, thin_border)
+for col, w in enumerate([12,8,8,12,10,26,18,22,10,22,13,12,10,14,16,14,14,12,10,8,12,12,12], 1):
+    ws4.column_dimensions[get_column_letter(col)].width = w
+
+# ── Save ──
+out = '7月核算结果/租赁任务考核_2026年7月_干预情况说明.xlsx'
+wb.save(out)
+
+print(f'\n✅ {out}')
+print(f'\n概况:')
+print(f'  全量: {total_vehicles}辆, {total_records}条')
+print(f'  干预: {interv_vehicles}辆, {interv_records}条')
+print(f'  达→未达标: {flip_total}条')
+print(f'  仪表盘里程: {total_tbox:.0f} (干预{interv_tbox:.0f})')
+print(f'  GPS里程: {total_actual:.0f} (干预{interv_actual:.0f})')
+print(f'  测试里程: {total_test:.0f} (干预{interv_test:.0f})')
diff --git a/replace_gps_jun.py b/replace_gps_jun.py
new file mode 100644
index 0000000..02b1ac6
--- /dev/null
+++ b/replace_gps_jun.py
@@ -0,0 +1,180 @@
+#!/usr/bin/env python3
+"""6月测试车辆：用GPS日里程替换仪表盘里程，写入源数据文件"""
+import openpyxl, pymysql, os
+from datetime import datetime, date
+from collections import defaultdict
+
+os.chdir('/Users/kkfluous/Downloads')
+
+# 1. 读取测试车辆清单
+fp = '/Users/kkfluous/Library/Containers/com.tencent.xinWeChat/Data/Documents/xwechat_files/wxid_1704407055112_3af8/temp/RWTemp/2026-07/65b41c10aa83cabb7a365b2b8437c1c7/6月测试车辆.xlsx'
+wb = openpyxl.load_workbook(fp, data_only=True)
+ws = wb.active
+rows = list(ws.iter_rows(min_row=1, values_only=True))
+test_plates = set()
+for row in rows[1:]:
+    for ci in [0, 3, 6]:
+        p = str(row[ci]).strip() if row[ci] else ''
+        if p and p != 'None':
+            test_plates.add(p)
+wb.close()
+print(f'测试车辆: {len(test_plates)}辆')
+
+# 2. 备份
+src = '租赁任务考核_2026年6月.xlsx'
+bak = '租赁任务考核_2026年6月_备份.xlsx'
+if not os.path.exists(bak):
+    import shutil
+    shutil.copy2(src, bak)
+    print(f'已备份: {bak}')
+
+# 3. 读源文件
+wb = openpyxl.load_workbook(src)
+ws = wb['业务考核视图']
+h = [c.value for c in next(ws.iter_rows(min_row=1, max_row=1))]
+recs = []
+for row in ws.iter_rows(min_row=2, values_only=True):
+    r = dict(zip(h, row))
+    r['row_num'] = len(recs) + 2
+    recs.append(r)
+
+# 测试车辆在源数据中的记录
+test_recs = [r for r in recs if r['车牌号'] in test_plates]
+print(f'测试车辆在6月源数据中的记录: {len(test_recs)}条 ({len(set(r["车牌号"] for r in test_recs))}辆)')
+
+# 4. 连数据库取GPS
+conn = pymysql.connect(host='101.133.130.65', port=3306, user='root', password='ln123!@#',
+                       database='hydrogen_energy', connect_timeout=30)
+cur = conn.cursor()
+
+def get_gps_km(plate, start_str, end_str):
+    """按考核起止日期汇总GPS日里程(km)"""
+    if not start_str or not end_str:
+        return None
+    try:
+        sd = datetime.strptime(str(start_str)[:10], '%Y-%m-%d').date()
+        ed = datetime.strptime(str(end_str)[:10], '%Y-%m-%d').date()
+    except:
+        return None
+    
+    cur.execute("""
+        SELECT dates, total_mileage FROM ln_vehicle_g7_mileage
+        WHERE plate_number = %s AND dates >= %s AND dates <= %s
+        ORDER BY dates
+    """, (plate, sd, ed))
+    rows = cur.fetchall()
+    if not rows:
+        return None
+    
+    # total_mileage is daily cumulative (meters). Get daily delta then sum.
+    # Actually check: is it daily km or cumulative? From last time: total_mileage/100000 = km
+    # Let's check if values are ascending (cumulative) or fluctuating (daily)
+    vals = [r[1] for r in rows]
+    is_cumulative = all(vals[i] <= vals[i+1] for i in range(len(vals)-1)) if len(vals) > 1 else False
+    
+    if is_cumulative:
+        # Daily delta: today - yesterday
+        deltas = []
+        prev = None
+        for v in vals:
+            if prev is not None and v > prev:
+                deltas.append(v - prev)
+            elif prev is None:
+                deltas.append(v)  # first day
+            prev = v
+        total_km = sum(deltas) / 100000
+    else:
+        total_km = sum(vals) / 100000
+    
+    return round(total_km, 1)
+
+# 批量查询: build plate→[(start,end,row_num)] map
+queries = defaultdict(list)
+for r in test_recs:
+    plate = r['车牌号']
+    start = r.get('考核开始日期')
+    end = r.get('考核结束日期')
+    queries[plate].append((start, end, r['row_num']))
+
+# Process each unique plate
+gps_cache = {}
+replaced = 0
+no_data = 0
+no_data_plates = set()
+
+for plate in sorted(queries.keys()):
+    # Get all GPS data for this plate in June for caching
+    cur.execute("""
+        SELECT dates, total_mileage FROM ln_vehicle_g7_mileage
+        WHERE plate_number = %s AND dates >= '2026-06-01' AND dates <= '2026-06-30'
+        ORDER BY dates
+    """, (plate,))
+    all_rows = cur.fetchall()
+    if not all_rows:
+        no_data += len(queries[plate])
+        no_data_plates.add(plate)
+        continue
+    
+    for start, end, row_num in queries[plate]:
+        km = get_gps_km(plate, start, end)
+        if km is None:
+            no_data += 1
+            no_data_plates.add(plate)
+        else:
+            gps_cache[row_num] = km
+
+conn.close()
+
+# 5. 更新源文件
+# Need to add '原TBOX里程' and '测试里程' columns if not exist
+# Check existing columns
+last_col_letter = openpyxl.utils.get_column_letter(ws.max_column)
+need_new_cols = '原TBOX里程(km)' not in h and '测试里程(km)' not in h
+
+if need_new_cols:
+    # Add two new columns
+    tbox_col = ws.max_column + 1
+    test_col = ws.max_column + 2
+    ws.cell(row=1, column=tbox_col, value='原TBOX里程(km)')
+    ws.cell(row=1, column=test_col, value='测试里程(km)')
+    # Find the actual行驶里程 column index
+    actual_idx = h.index('实际行驶里程(km)') + 1  # 1-indexed
+else:
+    tbox_idx = h.index('原TBOX里程(km)') + 1 if '原TBOX里程(km)' in h else None
+    test_idx = h.index('测试里程(km)') + 1 if '测试里程(km)' in h else None
+
+# Apply replacements
+for row_num, gps_km in gps_cache.items():
+    # Find column indices in current sheet
+    row_data = [c.value for c in ws[row_num]]
+    h_current = [c.value for c in ws[1]]
+    
+    actual_col = h_current.index('实际行驶里程(km)') + 1
+    old_val = ws.cell(row=row_num, column=actual_col).value
+    
+    if need_new_cols:
+        tbox_col = h_current.index('原TBOX里程(km)') + 1 if '原TBOX里程(km)' in h_current else ws.max_column - 1
+        test_col = h_current.index('测试里程(km)') + 1 if '测试里程(km)' in h_current else ws.max_column
+    else:
+        tbox_col = h_current.index('原TBOX里程(km)') + 1
+        test_col = h_current.index('测试里程(km)') + 1
+    
+    # Write original TBOX mileage
+    ws.cell(row=row_num, column=tbox_col, value=float(old_val) if old_val else 0)
+    # Write test mileage (TBOX - GPS)
+    tbox_val = float(old_val or 0)
+    test_km = round(tbox_val - gps_km, 1)
+    ws.cell(row=row_num, column=test_col, value=test_km)
+    # Replace actual mileage with GPS
+    ws.cell(row=row_num, column=actual_col, value=gps_km)
+    replaced += 1
+
+wb.save(src)
+wb.close()
+
+print(f'\n替换结果:')
+print(f'  成功替换: {replaced}条')
+print(f'  无GPS数据: {no_data}条 ({len(no_data_plates)}辆)')
+if no_data_plates:
+    print(f'  无数据车牌: {sorted(no_data_plates)}')
+print(f'\n已保存: {src}')
diff --git a/replace_gps_jun_new.py b/replace_gps_jun_new.py
new file mode 100644
index 0000000..14b1f8e
--- /dev/null
+++ b/replace_gps_jun_new.py
@@ -0,0 +1,169 @@
+#!/usr/bin/env python3
+"""Step 1: GPS里程替换 - 针对 租赁任务考核_2026年6月 (1).xlsx"""
+import openpyxl, pymysql, os, shutil
+from datetime import datetime
+from collections import defaultdict
+
+os.chdir('/Users/kkfluous/Downloads')
+
+# 36辆测试车辆（从上一次GPS替换留底提取）
+TEST_PLATES = {
+    '粤AGE5480','粤AGF4535','粤AGG3490','粤AGG4135','粤AGG4191',
+    '粤AGP2033','粤AGP3080','粤AGP3663','粤AGP3676','粤AGP5139',
+    '粤AGP5156','粤AGP5163','粤AGP5168','粤AGP5636','粤AGP5767',
+    '粤AGP5768','粤AGP5796','粤AGP9733','粤AGP9786','粤AGR0298',
+    '粤AGR1586','粤AGR3288','粤AGR5056','粤AGR5068','粤AGR5088',
+    '粤AGR6869','粤AGR6877','粤AGR6879','粤AGR6881','粤AGR8506',
+    '粤AGR8551','粤AGR8558','粤AGR8799','粤AGR9866','粤AGR9893',
+    '粤AGR9899',
+}
+
+SRC = '租赁任务考核_2026年6月 (1).xlsx'
+DST = '租赁任务考核_2026年6月.xlsx'
+BAK = '租赁任务考核_2026年6月_备份_0714.xlsx'
+
+# 0. 备份现有的6月文件
+if os.path.exists(DST):
+    shutil.copy2(DST, BAK)
+    print(f'已备份原6月文件 → {BAK}')
+
+# 1. 复制新源文件为目标文件
+shutil.copy2(SRC, DST)
+print(f'源文件: {SRC}')
+print(f'目标文件: {DST}')
+
+# 2. 读取目标文件
+wb = openpyxl.load_workbook(DST)
+ws = wb['业务考核视图']
+h = [c.value for c in next(ws.iter_rows(min_row=1, max_row=1))]
+print(f'列数: {len(h)}, 表头: {h}')
+
+recs = []
+for row in ws.iter_rows(min_row=2, values_only=True):
+    r = dict(zip(h, row))
+    r['row_num'] = len(recs) + 2
+    recs.append(r)
+
+test_recs = [r for r in recs if r['车牌号'] in TEST_PLATES]
+print(f'测试车辆记录: {len(test_recs)}条 ({len(set(r["车牌号"] for r in test_recs))}辆)')
+
+# 3. 连接数据库取GPS日里程
+conn = pymysql.connect(host='101.133.130.65', port=3306, user='root', password='ln123!@#',
+                       database='hydrogen_energy', connect_timeout=30)
+cur = conn.cursor()
+
+def get_gps_km(plate, start_str, end_str):
+    """汇总GPS日里程(km)"""
+    if not start_str or not end_str:
+        return None
+    try:
+        sd = datetime.strptime(str(start_str)[:10], '%Y-%m-%d').date()
+        ed = datetime.strptime(str(end_str)[:10], '%Y-%m-%d').date()
+    except:
+        return None
+
+    cur.execute("""
+        SELECT dates, total_mileage FROM ln_vehicle_g7_mileage
+        WHERE plate_number = %s AND dates >= %s AND dates <= %s
+        ORDER BY dates
+    """, (plate, sd, ed))
+    rows = cur.fetchall()
+    if not rows:
+        return None
+
+    vals = [r[1] for r in rows]
+    is_cumulative = all(vals[i] <= vals[i+1] for i in range(len(vals)-1)) if len(vals) > 1 else False
+
+    if is_cumulative:
+        deltas = []
+        prev = None
+        for v in vals:
+            if prev is not None and v > prev:
+                deltas.append(v - prev)
+            elif prev is None:
+                deltas.append(v)
+            prev = v
+        total_km = sum(deltas) / 100000
+    else:
+        total_km = sum(vals) / 100000
+
+    return round(total_km, 1)
+
+# 批量查询
+gps_cache = {}
+replaced = 0
+no_data = 0
+no_data_plates = set()
+
+for r in test_recs:
+    plate = r['车牌号']
+    km = get_gps_km(plate, r.get('考核开始日期'), r.get('考核结束日期'))
+    if km is None:
+        no_data += 1
+        no_data_plates.add(plate)
+    else:
+        gps_cache[r['row_num']] = km
+
+conn.close()
+print(f'GPS查询: 成功{len(gps_cache)}条, 无数据{no_data}条')
+
+# 4. 添加 原TBOX里程(km) 和 测试里程(km) 列
+if '原TBOX里程(km)' not in h:
+    tbox_col = ws.max_column + 1
+    test_col = ws.max_column + 2
+    ws.cell(row=1, column=tbox_col, value='原TBOX里程(km)')
+    ws.cell(row=1, column=test_col, value='测试里程(km)')
+    # 刷新表头
+    h_new = [c.value for c in next(ws.iter_rows(min_row=1, max_row=1))]
+else:
+    h_new = h
+
+# 找列索引
+actual_idx = h_new.index('实际行驶里程(km)') + 1
+tbox_idx = h_new.index('原TBOX里程(km)') + 1
+test_idx = h_new.index('测试里程(km)') + 1
+
+# 5. 写入GPS数据
+for row_num, gps_km in gps_cache.items():
+    old_val = ws.cell(row=row_num, column=actual_idx).value
+
+    # 写入原TBOX里程
+    ws.cell(row=row_num, column=tbox_idx, value=float(old_val) if old_val else 0)
+    # 写入测试里程
+    tbox_val = float(old_val or 0)
+    test_km = round(tbox_val - gps_km, 1)
+    ws.cell(row=row_num, column=test_idx, value=test_km)
+    # 替换实际行驶里程为GPS
+    ws.cell(row=row_num, column=actual_idx, value=gps_km)
+    replaced += 1
+
+# 6. 测试车辆标记黄色背景
+from openpyxl.styles import PatternFill
+yellow_fill = PatternFill(start_color='FFFF00', end_color='FFFF00', fill_type='solid')
+for row_num in gps_cache.keys():
+    for c in range(1, ws.max_column + 1):
+        ws.cell(row=row_num, column=c).fill = yellow_fill
+
+# 7. 重算完成率和是否达标
+DAYSM = {1:31,2:28,3:31,4:30,5:31,6:30,7:31,8:31,9:30,10:31,11:30,12:31}
+for r in recs:
+    rn = r['row_num']
+    actual = ws.cell(row=rn, column=actual_idx).value or 0
+    target = float(r.get('应考核里程(km)') or 0)
+    pct = round(actual / target * 100, 2) if target > 0 else 0
+    q = '达标' if actual >= target and target > 0 else '未达标'
+    # 找完成率列和是否达标列
+    rate_idx = h_new.index('完成率(%)') + 1
+    qualified_idx = h_new.index('是否达标') + 1
+    ws.cell(row=rn, column=rate_idx, value=pct)
+    ws.cell(row=rn, column=qualified_idx, value=q)
+
+wb.save(DST)
+wb.close()
+
+print(f'\n===== GPS替换完成 =====')
+print(f'替换成功: {replaced}条')
+print(f'无GPS数据: {no_data}条 ({len(no_data_plates)}辆)')
+if no_data_plates:
+    print(f'无数据车牌: {sorted(no_data_plates)}')
+print(f'已保存: {DST}')
diff --git a/yangchanji-plan-ab-compare.md b/yangchanji-plan-ab-compare.md
new file mode 100644
index 0000000..0c7c4af
--- /dev/null
+++ b/yangchanji-plan-ab-compare.md
@@ -0,0 +1,211 @@
+# 养馋记 · 物流台账闭环 · 全量对照总表
+
+> **Make 打开本页：** http://localhost:53817/?projectId=make&doc=yangchanji-plan-ab-compare.md  
+> **主库：** `knowledge/产品线1-项目1/04-产品方案/方案AB对照摘要.md`
+
+| 文档 | Make 打开 |
+|------|-----------|
+| **本页 · A/B 对照** | http://localhost:53817/?projectId=make&doc=yangchanji-plan-ab-compare.md |
+| 方案 A · 可对接 | http://localhost:53817/?projectId=make&doc=yangchanji-plan-a.md |
+| 方案 B · 不可对接 | http://localhost:53817/?projectId=make&doc=yangchanji-plan-b.md |
+| 数智中心评估汇总 | http://localhost:53817/?projectId=make&doc=yangchanji-ledger-eval.md |
+
+| 文档 | 用途 |
+|------|------|
+| 本页 | A/B 与共性一表看清 |
+| 方案 A | 可对接 API 的完整评估（第五、六点） |
+| 方案 B | 不可对接时的完整评估（第五、六点） |
+| 数智中心评估汇总 | 双路径合一稿 |
+
+---
+
+## 一、方案基本信息对照
+
+| 对照项 | 方案 A · 可对接 | 方案 B · 不可对接 |
+|--------|-----------------|-------------------|
+| 前提 | 养馋记 TMS 提供可用 API + 沙箱 | 无可用 API / 短期对不通 |
+| 数据进 OneOS | 对方 API 同步（Excel 兜底） | 导入模板 / 手工建单 + 主数据改造 |
+| 状态回传对方 | 可做（P1） | 本期不做（可导出） |
+| 特有开发 | 对接适配层 | 导入中心 + 字段改造 |
+| 可行性 | 有条件可行 | 有条件可行 |
+| 推荐场景 | 接口确认可用时默认 | 对不通，或先通闭环再切 A |
+| 版本 | V1.2（表格化） | V1.2（表格化） |
+
+---
+
+## 二、第五点八问 · A/B 对照
+
+| # | 评估问题 | 方案 A | 方案 B |
+|---|----------|--------|--------|
+| 1 | 技术上能否实现 | 有条件可行（依赖 API） | 有条件可行（不依赖 API） |
+| 2 | 数据/接口怎么来 | API：订单/路线/门店等；回传 P1 | 无接口；导入/录入同套业务对象 |
+| 3 | 我方开发什么 | 对接层 + 待派派单小程序计费对账 | 导入改造 + 待派派单小程序计费对账 |
+| 4 | 多长时间 | 约 **18～25 周 / 5～6.5 月** | 约 **20～27 周 / 5.5～7 月** |
+| 5 | 投入人员 | 产1研2测1 + 对方接口人 | 产1研2测1 + **运营导单强依赖** |
+| 6 | 额外费用 | 联调+小程序等；专线视情况 | 现金更低；**导单隐性成本中** |
+| 7 | 前置与风险 | API/沙箱/映射；字段不一致、审核延误 | 导单责任/缺口表；导单延误、数据不一致 |
+| 8 | 分期 | 调研→V0→V1→V2→利润后置 | 同左（V1 多 2～4 周） |
+
+---
+
+## 三、周期对照（已压缩调研+V0）
+
+| 阶段 | 方案 A | 方案 B | 备注 |
+|------|--------|--------|------|
+| 业务调研 | 1～2 周 | 1～2 周 | 与 V0 可并行收口 |
+| V0 门禁 | 约 1 周 | 约 1 周 | |
+| **调研+V0** | **约 2～3 周** | **约 2～3 周** | 原 3～6 周减半 |
+| V1 | 12～16 周 | 14～18 周 | B 含导入改造 |
+| V2 | 4～6 周 | 4～6 周 | |
+| 利润（后置） | 另 2～4 周 | 另 2～4 周 | 不进首批退出 |
+| **首批合计** | **约 5～6.5 个月** | **约 5.5～7 个月** | 含调研 |
+
+---
+
+## 四、业务时序与已拍板规则（A/B 相同）
+
+| 步骤 | 规则 |
+|------|------|
+| 1 | 数据进入 OneOS（A=API / B=导入或录入） |
+| 2 | 自动生成待派任务 |
+| 3 | 人工选车、选司机并派单 |
+| 4 | **派单后**司机小程序才可见任务 |
+| 5 | 正常情况：**当前运单未完成，不可接下一单** |
+| 6 | 办结运单后 → **才触发正式计费/出账** |
+| 7 | V2：成本归集 + 流水 + 合作方只读确认 |
+| 8 | 利润抽点在流水与应收应付之后 |
+
+| 已拍板项 | 结论 |
+|----------|------|
+| 首批范围 | V0+V1+V2；V3（用友等）不计入 |
+| 计费 | 多模型引擎；每客户只绑一种 |
+| 对账深度 | 只读 + 确认（无驳回/补凭证） |
+| 司机端 | 微信小程序，V1 必达 |
+| 氢/ETC | 可先运费；V2 必须做完 |
+| 编制 | 产品1 + 研发2 + 测试1 |
+
+| 待确认项 | 说明 |
+|----------|------|
+| 走 A / B / 先 B 后 A | 调研结束书面选定 |
+| 车司机是否固定绑定 | 运营确认 |
+| 异常是否允许代办结 | 运营确认 |
+| 规则包书面试算样例 | 业务+财务 |
+| 利润具体参数 | 后置再锁 |
+
+---
+
+## 五、数据对象对照（业务要什么 × 怎么拿到）
+
+| 数据对象 | 用途 | 方案 A 获取 | 方案 B 获取 | 优先级 |
+|----------|------|-------------|-------------|--------|
+| 订单 | 生成待派 | 对方 API | 导入/手工建单 | P0 |
+| 路线/经停 | 派单与司机点序 | 对方 API | 导入/录入 | P0 |
+| 门店主数据 | 主数据与对账 | API 或首导+增量 | 批量导入+日常维护 | P0 |
+| 货物/物料 | 温层货量/计价 | API（汇总或明细） | 随单导入或汇总填 | P0/按需 |
+| 运费要素 | 办结后计费 | 订单内或规则侧 | 导入列或规则侧 | P0 |
+| 车/司机 | 人工派单 | 我方主数据为主 | 同左 | P0 |
+| 状态回传 | 告知对方 | OneOS→对方 API | 不做接口（可导出） | A:P1 / B:— |
+| 氢费/ETC | V2 成本 | 我方内部系统 | 同左 | V2 |
+
+---
+
+## 六、我方功能对照
+
+| 功能模块 | 方案 A | 方案 B | 阶段 |
+|----------|--------|--------|------|
+| 对接适配（鉴权/同步/映射/幂等/重试/差异日志） | ✅ | ❌（可不建） | V0～V1 |
+| 主数据/单据字段改造 | 按需 | ✅ 重点 | V0～V1 |
+| 导入中心（模板/校验/重导） | 仅兜底 | ✅ 重点 | V1 |
+| 手工建单/维护 | 辅 | ✅ | V1 |
+| 待派任务池 | ✅ | ✅ | V1 |
+| 人工调度台 | ✅ | ✅ | V1 |
+| 司机微信小程序 | ✅ | ✅ | V1 |
+| 办结后计费出账 | ✅ | ✅ | V1 |
+| 成本归集 | ✅ | ✅ | V2 |
+| 流水/应收应付/只读确认 | ✅ | ✅ | V2 |
+| 利润抽点 | 后置 | 后置 | V2 后 |
+| 用友 | 不计入 | 不计入 | V3 |
+
+---
+
+## 七、人员与费用对照
+
+| 角色/费用项 | 方案 A | 方案 B |
+|-------------|--------|--------|
+| 产品 | 1 | 1 |
+| 研发 | 2 | 2 |
+| 测试 | 1 | 1 |
+| 养馋记配合 | 接口人强 | 台账/导出样例弱 |
+| 运营配合 | 调研+双跑 | **导单责任人+日常导单（强）** |
+| 对方 API 联调费 | 可能有 | 无 |
+| 小程序认证 | 必需 | 必需 |
+| 短信 | 可选 | 可选 |
+| 氢/ETC 源 | V2 必需 | V2 必需 |
+| 独立服务器 | 通常否 | 通常否 |
+| 隐性成本 | 联调等待 | **持续导单与主数据质量** |
+
+---
+
+## 八、第六点确认事项 · A/B 对照
+
+| # | 原文件确认项 | 方案 A 状态 | 方案 B 状态 |
+|---|--------------|-------------|-------------|
+| 1 | 标准接口还是 Excel | 假定 **API**；Excel 兜底 | 假定 **无 API**；Excel/模板为主 |
+| 2 | 能提供哪些字段 | 待映射表收口 | 按业务 P0 定导入列 |
+| 3 | 状态能否实时回传 | 目标 P1；调研确认是否准实时 | **本期不做**系统回传 |
+| 4 | 氢费/ETC 从哪来 | 我方内部；V2 | 同左 |
+| 5 | 司机端形态 | **已定小程序** | **已定小程序** |
+| 6 | 车司机是否绑定 | 待运营确认 | 待运营确认 |
+| 7 | 利润怎么配 | 后置；按客户配 | 同左 |
+| 8 | 对账深度 | **已定只读+确认** | **已定只读+确认** |
+| 9 | 是否接用友 | **本期不做** | **本期不做** |
+
+---
+
+## 九、分期退出标准对照
+
+| 期次 | 方案 A 退出标准 | 方案 B 退出标准 |
+|------|-----------------|-----------------|
+| 调研 | 确认可对接；字段草案+原型范围 | 确认走 B（或先 B）；缺口表+导单责任+原型范围 |
+| V0 | 映射确认、规则样例、小程序账号、沙箱就绪 | 导入模板、字段缺口确认、规则样例、小程序账号就绪 |
+| V1 | 双跑运费可解释；小程序可用；办结出账；未完成不可接下一单 | 导入可生成待派；其余同 A |
+| V2 | 成本进项目；双方月确认一版 | 同左（对内或合作方账号） |
+| 利润 | 后置 | 后置 |
+
+| 名词 | 含义（白话） |
+|------|----------------|
+| 退出标准 | 本期做到什么才算结束、可进下一期 |
+| 双跑出账可解释（V1） | 运费应收侧能讲清；完整应付成本在 V2 |
+| 映射确认 | A=对方字段↔OneOS；B=导入列↔OneOS |
+
+---
+
+## 十、决策日志 · 养馋记台账相关（已拍板对照）
+
+| 日期 | 议题 | 结论 |
+|------|------|------|
+| 2026-08-10 | 首批范围 | 批准 V0+V1+V2；V3 不计入 |
+| 2026-08-10 | 计费模型 | 多模型引擎；每客户只绑一种 |
+| 2026-08-10 | 对账深度 | 只读+确认 |
+| 2026-08-10 | 司机端 | 微信小程序；V1 必达 |
+| 2026-08-10 | 氢/ETC | 可先运费；V2 必做完 |
+| 2026-08-10 | 利润 | 流水与应收应付之后 |
+| 2026-08-10 | 编制 | 产1研2测1 |
+| 2026-08-10 | 时序 | 派单后可见；办结后计费 |
+| 2026-08-10 | 接单约束 | 未完成不可接下一单 |
+| 2026-08-11 | 数据路径 | A API / B 导入改造 |
+| 2026-08-11 | 周期 | 调研+V0≈2～3周；A≈5～6.5月；B≈5.5～7月 |
+
+---
+
+## 十一、评估结果一览（并排）
+
+| 评估结果项 | 方案 A | 方案 B |
+|------------|--------|--------|
+| 可行性 | 有条件可行 | 有条件可行 |
+| 数据策略 | 对方 API 主路径 | OneOS 改造+导入/维护 |
+| 周期 | 约 5～6.5 个月 | 约 5.5～7 个月 |
+| 人员 | 产1研2测1 | 产1研2测1+运营强依赖 |
+| 费用感 | 低～中 | 现金低、运营成本中 |
+| 首批 | 调研→V0→V1→V2 | 同左 |
+| 利润/用友 | 后置 / 不计入 | 同左 |
diff --git a/养馋记物流派单与对账模块_管理层汇报.md b/养馋记物流派单与对账模块_管理层汇报.md
new file mode 100644
index 0000000..debbcd8
--- /dev/null
+++ b/养馋记物流派单与对账模块_管理层汇报.md
@@ -0,0 +1,97 @@
+# 养馋记物流派单与对账模块建设评估
+## 管理层汇报版
+
+> 本次仅进行方案与工作量评估，暂不进入开发。  
+> 目标：养馋记订单及路线数据接入 **OneOS**，形成“派单—费用归集—利润核算—月度对账”闭环。
+
+---
+
+## 一、结论与请决
+
+### 结论
+
+- **项目可做，但以 API、费用口径和成本数据可用为前提。**
+- **首期采用“系统辅助 + 人工派单”**，不做全自动排班。
+- 按“三阶段”推进：先跑通业务与 T+1 暂算盈亏，再完成对账闭环，最后建设 AI 智能调度。
+
+### 请决事项
+
+1. 同意第一阶段按 **1 个月 MVP** 推进；
+2. 同意以养馋记 API 为主、Excel/人工建单为兜底；
+3. 同意第一阶段配置产品 1 人、研发 2 人、测试 1 人；
+4. 同意 API 门禁、字段清单、费用试算样例和 OneOS 成本匹配规则确认后，正式启动开发。
+
+---
+
+## 二、分阶段投入总表
+
+| 阶段 | 核心交付 | 周期 | 人员配置 | 阶段人力成本 | 累计人力成本 |
+| --- | --- | ---: | --- | ---: | ---: |
+| **启动门禁** | API/字段确认、业务调研、计费试算、OneOS 成本数据匹配、原型与测试环境 | 约 1 周 | 产品 1、研发 2；测试按需 | 纳入第一阶段 | — |
+| **第一阶段：MVP** | 订单接入、物流台账、待派任务、人工派车/派司机、司机端基础执行、OneOS 氢费/ETC 归集、逐单及 T+1 暂算盈亏 | **约 1 个月** | 产品 1、研发 2、测试 1 | **96,600 元** | **96,600 元** |
+| **第二阶段：业务与对账闭环** | 改单取消、异常办结、完整费用与车辆成本、计费规则、利润精算、明细账/汇总账、月度对账、合作方在线对账、GPS 异常管控 | **2—3 个月** | 产品 1、研发 2、测试 1 | **206,400—322,200 元** | **303,000—418,800 元** |
+| **第三阶段：AI 智能调度试点** | 车辆/司机推荐、串单串点、复跑优化、车型装载匹配、空驶/绕行/无效 ETC 优化 | 数据积累 3—6 个月；试点开发 3—4 个月 | 产品 1、研发/算法 2—3、测试 1 | **475,200—633,600 元** | **778,200—1,052,400 元** |
+
+### 成本口径
+
+| 成本项 | 口径 |
+| --- | --- |
+| 产品、研发 | 1,500 元/人天 |
+| 测试 | 1,200 元/人天 |
+| 第一、二阶段满负荷日成本 | 5,700 元/工作日 |
+| 当前建议审批范围 | **第一阶段 + 第二阶段：303,000—418,800 元** |
+| 含 AI 试点的远期总投入 | **778,200—1,052,400 元**，待数据准备完成后单独立项 |
+
+> 第二、三阶段为评估值；正式报价按 API 成熟度、费用规则数量、对账权限范围及实际人天核定。
+
+---
+
+## 三、各阶段能做什么
+
+| 能力 | 第一阶段 | 第二阶段 | 第三阶段 |
+| --- | --- | --- | --- |
+| 养馋记订单、路线接入 | ✔ | 完善异常与状态回传 | — |
+| 待派任务、人工派单 | ✔ | 完善拆单、换车换人、异常流程 | 智能推荐 |
+| 司机端执行与凭证 | ✔ | 完善异常与闭环 | — |
+| OneOS 氢费、ETC 归集 | ✔（基础归集） | 完整费用、车辆成本和精算 | 成本优化建议 |
+| 逐单利润 | ✔（暂算） | ✔（精算、可对账） | 优化决策 |
+| 月度账单、合作方在线对账 | — | ✔ | — |
+| AI 智能调度 | — | 数据积累 | ✔（试点） |
+
+### 第一阶段验收标准
+
+> **养馋记订单接入 → 待派任务 → 人工派车/派司机 → 司机执行与办结 → OneOS 氢费/ETC 归集 → 逐单暂算 → T+1 盈亏查看**
+
+首期不包含：完整月度对账、合作方确认/驳回、用友接口、全自动派单及 AI 调度。
+
+---
+
+## 四、成本事项
+
+| 类别 | 具体事项 | 是否纳入上述人力成本 |
+| --- | --- | --- |
+| 人力成本 | 产品、研发、测试投入 | **已纳入** |
+| API 与联调 | 养馋记接口服务、联调支持、可能的信息服务费 | 未纳入，待确认 |
+| 平台服务 | 小程序认证、短信/消息、地图、GPS 服务 | 未纳入，按实际使用核定 |
+| 基础设施 | 云服务器、数据库、对象存储、独立部署及运维 | 未纳入，按部署方案核定 |
+| 财务/第三方系统 | 用友等系统后续接口 | 未纳入，本期不做 |
+
+---
+
+## 五、启动前门禁与主要风险
+
+| 门禁事项 | 不确认的影响 |
+| --- | --- |
+| 养馋记 API 文档、鉴权、沙箱、字段和同步时效 | 首期范围和周期无法锁定 |
+| 新单、改单、取消单、拆单及状态回传规则 | 任务与派单逻辑不完整 |
+| 运费、抽点、杂费、车辆成本的试算样例 | 利润和对账结果可能有争议 |
+| OneOS 氢费、ETC 的来源、时效和运单匹配规则 | 只能形成暂算，无法精算 |
+| 合作方对账权限：只读、确认、驳回、补凭证 | 第二阶段在线对账无法闭环 |
+
+---
+
+## 六、建议决策
+
+1. **本次先批准第一阶段 MVP，预算 96,600 元，周期约 1 个月。**
+2. 第一阶段验收后，再按实际业务量和对账规则确认第二阶段 2—3 个月投入。
+3. AI 智能调度不纳入当前开发承诺；待业务闭环稳定、累计 3—6 个月真实数据后，以试点方式单独立项。
diff --git a/养馋记物流派单与对账模块_管理层汇报_精简版.md b/养馋记物流派单与对账模块_管理层汇报_精简版.md
new file mode 100644
index 0000000..e719ae2
--- /dev/null
+++ b/养馋记物流派单与对账模块_管理层汇报_精简版.md
@@ -0,0 +1,26 @@
+# 养馋记物流派单与对账模块建设评估
+
+## 一、结论与请决
+
+- **项目可做**：以养馋记 API、费用规则及 OneOS 成本数据可用为前提。
+- **首期策略**：系统辅助、人工派单；暂不做全自动与 AI 调度。
+- **本次请决**：同意先启动第一阶段 MVP，接口与试算样例确认后正式开发。
+
+## 二、分阶段投入
+
+| 阶段 | 主要交付 | 周期 | 人力成本 |
+| --- | --- | --- | --- |
+| 第一阶段：MVP | 订单接入、人工派单、司机执行、氢费/ETC 归集、逐单及 T+1 暂算盈亏 | 1 个月 | **114,000 元** |
+| 第二阶段：对账闭环 | 异常处理、费用精算、月度账单、合作方在线对账 | 1—2 个月 | **114,000—228,000 元**；累计 **228,000—342,000 元** |
+| 第三阶段：AI 试点 | 串单/串点、复跑、车型匹配及成本优化建议 | 数据积累 2—3 个月；试点 1—2 个月 | **114,000—288,000 元**；远期累计 **342,000—630,000 元** |
+
+## 三、成本口径
+
+- 产品、研发：**1,500 元/人天**；测试：**1,200 元/人天**。
+- 第一、二阶段配置：产品 1 人、研发 2 人、测试 1 人；按 20 个工作日/月计算，日成本 **5,700 元**。
+- AI 试点配置：产品 1 人、研发/算法 2—3 人、测试 1 人；数据积累期随前两阶段业务运行完成，不单独计入开发人力。
+- 上述为人力成本；API 联调、小程序认证、短信/地图/GPS、云资源及独立部署按实际方案另计。
+
+## 四、启动门禁
+
+确认 API 与字段、订单状态规则、费用试算样例及 OneOS 氢费/ETC 匹配规则后启动。
diff --git a/租赁任务考核_2026年4月.xlsx b/租赁任务考核_2026年4月.xlsx
new file mode 100644
index 0000000..64e0308
Binary files /dev/null and b/租赁任务考核_2026年4月.xlsx differ
diff --git a/租赁任务考核_2026年4月_干预情况说明.xlsx.xlsx b/租赁任务考核_2026年4月_干预情况说明.xlsx.xlsx
new file mode 100644
index 0000000..efee37f
Binary files /dev/null and b/租赁任务考核_2026年4月_干预情况说明.xlsx.xlsx differ
diff --git a/租赁任务考核_2026年5月.xlsx b/租赁任务考核_2026年5月.xlsx
new file mode 100644
index 0000000..d38f9fc
Binary files /dev/null and b/租赁任务考核_2026年5月.xlsx differ
diff --git a/租赁任务考核_2026年5月_原版备份.xlsx b/租赁任务考核_2026年5月_原版备份.xlsx
new file mode 100644
index 0000000..2471c9e
Binary files /dev/null and b/租赁任务考核_2026年5月_原版备份.xlsx differ
diff --git a/租赁任务考核_2026年6月 (1).xlsx b/租赁任务考核_2026年6月 (1).xlsx
new file mode 100644
index 0000000..11cfe00
Binary files /dev/null and b/租赁任务考核_2026年6月 (1).xlsx differ
diff --git a/租赁任务考核_2026年6月.xlsx b/租赁任务考核_2026年6月.xlsx
new file mode 100644
index 0000000..aa6d92c
Binary files /dev/null and b/租赁任务考核_2026年6月.xlsx differ
diff --git a/租赁任务考核_2026年6月_GPS替换后留底.xlsx b/租赁任务考核_2026年6月_GPS替换后留底.xlsx
new file mode 100644
index 0000000..499ec2f
Binary files /dev/null and b/租赁任务考核_2026年6月_GPS替换后留底.xlsx differ
diff --git a/租赁任务考核_2026年6月_备份.xlsx b/租赁任务考核_2026年6月_备份.xlsx
new file mode 100644
index 0000000..5a8edd2
Binary files /dev/null and b/租赁任务考核_2026年6月_备份.xlsx differ
diff --git a/租赁任务考核_2026年6月_备份_0714.xlsx b/租赁任务考核_2026年6月_备份_0714.xlsx
new file mode 100644
index 0000000..ba7fa61
Binary files /dev/null and b/租赁任务考核_2026年6月_备份_0714.xlsx differ
diff --git a/租赁任务考核_2026年6月_备份_0715.xlsx b/租赁任务考核_2026年6月_备份_0715.xlsx
new file mode 100644
index 0000000..f8cf674
Binary files /dev/null and b/租赁任务考核_2026年6月_备份_0715.xlsx differ
diff --git a/租赁任务考核_2026年6月_干预情况说明.xlsx b/租赁任务考核_2026年6月_干预情况说明.xlsx
new file mode 100644
index 0000000..d652542
Binary files /dev/null and b/租赁任务考核_2026年6月_干预情况说明.xlsx differ
diff --git a/租赁任务考核_2026年6月_干预情况说明_0714.xlsx b/租赁任务考核_2026年6月_干预情况说明_0714.xlsx
new file mode 100644
index 0000000..f94c972
Binary files /dev/null and b/租赁任务考核_2026年6月_干预情况说明_0714.xlsx differ
diff --git a/租赁任务考核_2026年7月 (1).xlsx b/租赁任务考核_2026年7月 (1).xlsx
new file mode 100644
index 0000000..a692a4c
Binary files /dev/null and b/租赁任务考核_2026年7月 (1).xlsx differ
diff --git a/租赁任务考核_2026年7月.xlsx b/租赁任务考核_2026年7月.xlsx
new file mode 100644
index 0000000..8daa45e
Binary files /dev/null and b/租赁任务考核_2026年7月.xlsx differ
diff --git a/里程任务考核_1月核算.xlsx b/里程任务考核_1月核算.xlsx
index 0f5f87f..56c080a 100644
Binary files a/里程任务考核_1月核算.xlsx and b/里程任务考核_1月核算.xlsx differ
diff --git a/里程任务考核_2月核算.xlsx b/里程任务考核_2月核算.xlsx
index ac7f4a6..68e7f59 100644
Binary files a/里程任务考核_2月核算.xlsx and b/里程任务考核_2月核算.xlsx differ
diff --git a/里程任务考核_3月核算.xlsx b/里程任务考核_3月核算.xlsx
index f75e7b0..e55c2b1 100644
Binary files a/里程任务考核_3月核算.xlsx and b/里程任务考核_3月核算.xlsx differ
diff --git a/里程任务考核_4月核算.xlsx b/里程任务考核_4月核算.xlsx
new file mode 100644
index 0000000..f96ccc6
Binary files /dev/null and b/里程任务考核_4月核算.xlsx differ
diff --git a/里程任务考核_5月核算.xlsx b/里程任务考核_5月核算.xlsx
new file mode 100644
index 0000000..674b837
Binary files /dev/null and b/里程任务考核_5月核算.xlsx differ
diff --git a/里程任务考核_6月核算.xlsx b/里程任务考核_6月核算.xlsx
new file mode 100644
index 0000000..a5c02af
Binary files /dev/null and b/里程任务考核_6月核算.xlsx differ
diff --git a/里程考核绩效核算/核算结果/里程任务考核_1月核算.xlsx b/里程考核绩效核算/核算结果/里程任务考核_1月核算.xlsx
deleted file mode 100644
index 0f5f87f..0000000
Binary files a/里程考核绩效核算/核算结果/里程任务考核_1月核算.xlsx and /dev/null differ
diff --git a/里程考核绩效核算/核算结果/里程任务考核_2月核算.xlsx b/里程考核绩效核算/核算结果/里程任务考核_2月核算.xlsx
deleted file mode 100644
index ac7f4a6..0000000
Binary files a/里程考核绩效核算/核算结果/里程任务考核_2月核算.xlsx and /dev/null differ
diff --git a/里程考核绩效核算/核算结果/里程任务考核_3月核算.xlsx b/里程考核绩效核算/核算结果/里程任务考核_3月核算.xlsx
deleted file mode 100644
index f75e7b0..0000000
Binary files a/里程考核绩效核算/核算结果/里程任务考核_3月核算.xlsx and /dev/null differ
diff --git a/里程考核绩效核算/考核依据/车辆里程考核与奖金发放规则（V.1.2）.docx b/里程考核绩效核算/考核依据/车辆里程考核与奖金发放规则（V.1.2）.docx
deleted file mode 100644
index 0855d43..0000000
Binary files a/里程考核绩效核算/考核依据/车辆里程考核与奖金发放规则（V.1.2）.docx and /dev/null differ
diff --git a/里程考核绩效核算/里程考核核算.zip b/里程考核绩效核算/里程考核核算.zip
deleted file mode 100644
index 861b105..0000000
Binary files a/里程考核绩效核算/里程考核核算.zip and /dev/null differ