Compare commits
| Author | SHA1 | Date | |
|---|---|---|---|
 Jesse_ChenandClaude Opus 5.5
|
c0f059125b | ||
|
Jesse_ChenandClaude Opus 5.5
|
f1a169cb93 | ||
|
Jesse_ChenandClaude Opus 5.5
|
564eede340 |
@@ -1,5 +1,16 @@
|
||||
# 印度占星 Skill 更新日志
|
||||
|
||||
## 2026-10-01 — 普通对话进度栏不再提前打勾
|
||||
|
||||
- 发问后进度栏不再一下子出现四行「先回答你问的这件事 / 把判断钉在盘上 / 时间窗口怎么看 / 这周可以做什么」并全部打勾（那时盘还没算完）。现在只显示真实发生的步骤：读取分析方法 → 计算本命盘 → 正在写（按正文实际写到的标题）→ 完成；打开历史对话也一样（BUG-1145）。
|
||||
- Skill 不 bump；不改数据库结构；服务端事件不变。
|
||||
|
||||
## 2026-10-01 — 生时校正不再答两三题就提前结束；默认按「哪道题最能分开盘型」出题
|
||||
|
||||
- 修复：轮到「健康」方面的选择题时，这道题一直写不进去，系统就直接给出结果，后面还能问的题都不问了。修好后平均每次多问约 2 题（前后 10 分钟：2.8 → 4.6 题；前后 30 分钟：3.6 → 5.4 题），D9 / D10 判对的例数明显上升，没有一例把真实时间排除在外（BUG-1143）。
|
||||
- 「按盘型挑题」默认开启（范围在前后 30 分钟以内时）：优先问最能区分 D9 / D10 上升的题。77 例回放：前后 30 分钟 D9 判对 46 → 51 例、D10 52 → 55 例（BUG-1144）。
|
||||
- Skill 版本不变；不改数据库结构。
|
||||
|
||||
## 2026-10-01 — 普通对话写回答最多可以等 5 分钟（原来 70 秒）
|
||||
|
||||
- 推理较重的模型（例如 deepseek-v4-pro）写长回答不再在最后一节被截断；5 分钟只用来防模型卡死，不限制回答长度（BUG-1142）。
|
||||
|
||||
@@ -15347,3 +15347,48 @@
|
||||
- 相关记录：BUG-1051、BUG-1053、BUG-944。
|
||||
- 复发自：无
|
||||
- 修复版本：`codex/consult-answer-clock-20261001`
|
||||
|
||||
## BUG-1143 | 健康类区分选择题写不进焦点表，流程静默提前交付
|
||||
|
||||
- 状态：resolved（2026-10-01 Claude 直接执行；77 例真库持久回放修前修后对比；未部署时以部署后 health 为准）
|
||||
- 首次发现 / 最近更新：2026-10-01 / 2026-10-01
|
||||
- 影响面：`frontend/src/lib/rectification-agentic/v9/server-focus.ts`（区分类焦点写 `targetDomain`）、`answer-choice.ts::persistFocusAfterChoice`（写入失败被吞）、所有生时校正用户。
|
||||
- 用户现象：答完两三道选择题就直接出交付卡，后面还有家人、感情、搬迁等可答的题却不再问；与「按盘型挑题」开关无关。
|
||||
- 触发条件：任一轮的区分题落在健康领域（引擎领域名 `health_pressure`）。
|
||||
- 根因：区分类焦点把计划层领域名原样写入 `agentic_rectification_conversation_focuses.target_domain`；该列检查约束只收 `education/career/relationship/relocation/finance/health/family/other`。写入抛 `target_domain_check` 违约，被 `persistFocusAfterChoice` 的 catch 吞掉（日志仅 `retry next focus failed reason=tool_failed`），流程落到 exhaustion 出口 `terminalNote` 交付。采集类焦点早经 `persistableFocusDomain` 映射，09-13 `530f260f`（BUG-661～663）的区分类写入分支漏掉。假数据库测试不校验约束，所以一直没测出来。
|
||||
- 修复：新增 `focusTargetDomain`，所有意图的焦点写入都经 `persistableFocusDomain`；`method-followup.ts` 三处焦点与探针的领域比较改走 `sameCollectDomain`（BUG-672 规则）；不改数据库约束。
|
||||
- 验证：`frontend/tests/rectification-focus-target-domain-20261001.test.ts`（引擎全部领域名 × 四种意图映射后都在迁移的约束清单内；健康区分焦点存 `health` 且与 `health_pressure` 探针同域；源码合同禁止原样写领域名与 `probe.domain === focus.targetDomain`）。真库（PostgreSQL 17 替身）持久回放 77 例 × ±10/±30（`docs/research/varga_resolution_persisted_onoff_{before,after}_fix_2026_10_01.json`）：平均问题数 ±10 2.82→4.56、±30 3.58→5.43；问满 6 题 21→50、30→69；默认顺序下 D9 / D10 头段命中 ±10 65→70 / 64→65、±30 38→46 / 39→52；真值段保留全部 76/76；修后日志 `retry next focus failed` 0 条。
|
||||
- 防复发：焦点写入只能经 `focusTargetDomain`；引擎领域与表约束的对照由测试逐值锁定。
|
||||
- 相关记录：BUG-672（同族：健康 / 职业别名）、BUG-586、BUG-661～663、BUG-1144。
|
||||
- 复发自：BUG-672（别名未经归并，写入侧新分支漏掉）。
|
||||
- 修复版本：`codex/rectification-segment-order-20261001`。
|
||||
|
||||
## BUG-1144 | 「按盘型挑题」默认关闭
|
||||
|
||||
- 状态：resolved（2026-10-01；默认开启窗口 ≤ 61 分钟）
|
||||
- 首次发现 / 最近更新：2026-10-01 / 2026-10-01
|
||||
- 影响面：`core/segment-probe-order.ts`、`v9/score-persist.ts`。
|
||||
- 用户现象：盘型口径实现（BUG-1116）把按段选题做成环境变量 `RECTIFICATION_SEGMENT_ORDER=on` 才启用，生产默认按引擎信息增益出题，分盘头段命中低于研究「按段」列。
|
||||
- 根因：实现单允许 T5 延后；执行方因真库持久回放未跑通而默认关闭。
|
||||
- 修复：`segmentOrderEnabledFor(windowMinutes, flag)`：默认窗口 ≤ 61 分钟开启，`off` 关闭，`on` 为研究包络。
|
||||
- 验证：修复 BUG-1143 后的 77 例真库持久回放，开 vs 关：±10 D9 70→71、D10 65→67；±30 D9 46→51、D10 52→55；真值段保留四格全部 76/76。修 BUG-1143 前 ±30 开启曾丢 3 例真值段（`…before_fix…json`），根因即 BUG-1143（开启后健康题被排到第一位，更早撞上写入失败）。`frontend/tests/rectification-segment-order-default-20261001.test.ts`。
|
||||
- 防复发：放宽到 > 61 分钟前须用同一回放脚本在对应窗口重跑并逐格核对真值段保留。
|
||||
- 相关记录：BUG-1116、BUG-1143、BUG-1105。
|
||||
- 复发自：无
|
||||
- 修复版本：`codex/rectification-segment-order-20261001`。
|
||||
|
||||
## BUG-1145 | 普通对话进度栏提前打勾：回答提纲在算盘前就显示为「已完成」
|
||||
|
||||
- 状态：resolved（2026-10-01 Claude 直接执行；单测复现修前失败、修后通过；未部署时以部署后 health 为准）
|
||||
- 首次发现 / 最近更新：2026-10-01 / 2026-10-01
|
||||
- 影响面：`frontend/src/lib/consultation-run-timeline.ts`（普通对话进度栏的实时与历史重建共用同一个 reducer）；生时校正进度栏不用这个 reducer，不受影响。
|
||||
- 用户现象：发问后，「先回答你问的这件事」「把判断钉在盘上」「时间窗口怎么看」「这周可以做什么」四行立刻全部打勾，而这时「正在计算本命盘…」还在转，正文一个字都没写。
|
||||
- 触发条件：任何本命普通对话首轮（通用 / 每日 / 申报时段的计划节同理）。
|
||||
- 根因：服务端在第一个流块就下发回答提纲（`think.plan` + 每节一条 `thinking.section`）；reducer 把每节做成一行 live 的 think 行，而 `completeLiveThink` 在下一个事件（下一节、`tool.started`）到来时把所有 live think 行标成完成。提纲是「回答将怎么写」，不是已走过的步骤，却按步骤渲染，于是算盘一开始四行就假完成。自 09-18 进度栏设计（`94c1e81f`）起即存在；BUG-1132 之后回答按人分段、「盘上依据」「时间怎么看」可省，提纲与真实正文进一步对不上。
|
||||
- 修复：产品 10-01 选 B——提纲不再成行。`think.plan` 不改状态；`thinking.section` 只把领域与对照项暂存为计算行的明细（`calculateHints`），在计算行出现时补上（实时与历史一致）；正文开始（`answer.delta`）时计算行收口为完成；历史重建的思考文本改走 `thinking.delta` 行，不再挂在提纲行上。服务端仍下发 `think.plan` / `thinking.section`（公共事件合同与存档字段不变，供 `applyThinkingSectionProgress` 等使用），客户端只是不再把它们画成行。
|
||||
- 验证：`frontend/tests/consultation-run-timeline.test.ts` 新增三条（算盘前提纲不成行、只有「读取分析方法」完成；整轮结束无提纲行、无 live 行；历史重建与实时同形且计算行仍有明细），三条在修前代码上失败、修后通过；改写一条既有断言（三栏见测试注释）。
|
||||
- 防复发：进度栏只画真实发生的步骤（方法、计算、写作、模型推理）；回答提纲若要再上屏，必须由正文标题驱动、不能用「下一事件到来即完成」的规则。
|
||||
- 相关记录：BUG-942～944（进度 / 思考 / 正文分通道）、BUG-1074（发出即显示排队行）、BUG-1132（按人分段）。
|
||||
- 复发自：无
|
||||
- 修复版本：`codex/consult-timeline-no-plan-20261001`。
|
||||
|
||||
|
||||
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
@@ -0,0 +1,34 @@
|
||||
# PROGRESS · 普通对话进度栏去掉回答提纲行（2026-10-01）
|
||||
|
||||
任务书：`docs/tasks/TASK-consult-timeline-no-plan-20261001.md`；BUG-1145。
|
||||
|
||||
## 实现
|
||||
|
||||
- `frontend/src/lib/consultation-run-timeline.ts`：
|
||||
- `think.plan` 直接返回原状态（不再生成提纲行）。
|
||||
- `thinking.section` 只把领域与对照项累积到新增的 `calculateHints`，经 `applyCalculateHints` 补到计算行；`tool.started` 建计算行时也补上（提纲在算盘前下发，原 `enrichCalculate` 在实时流里因计算行尚不存在而落空，历史重建里才有明细；现在两边一致）。
|
||||
- `answer.delta` 同时把计算行收口为完成（原来由 `thinking.section` 顺手收口）。
|
||||
- `consultationTimelineFromSettled`：有思考文本就发一条 `thinking.delta`，删掉「把思考文本挂到最后一条提纲行」的补丁。
|
||||
- 服务端 `think.plan` / `thinking.section` 照发（公共事件合同、`consultation-agentic-runtime` 测试锁定 `think.plan` 存在；客户端已不再用它画行）。`thinkPlanSteps` 仍被服务端与测试使用，未删。
|
||||
|
||||
## 改动的断言（原值 / 新值 / 原因）
|
||||
|
||||
| 测试 | 原值 | 新值 | 原因 |
|
||||
| --- | --- | --- | --- |
|
||||
| `consultation-run-timeline.test.ts` 首条（改名为 “…then the first write (no outline row)”） | 行种类 `["method","calculate","think","write"]`，第 3 行「先回答你问的这件事」已完成 | 行种类 `["method","calculate","write"]`，不存在该标签行；写作行下标 3→2 | BUG-1145：提纲不是步骤，产品选 B 去掉 |
|
||||
|
||||
新增三条：算盘前提纲不成行且只有方法行完成；整轮结束无 think 行、无 live 行；历史重建同形且计算行有明细。三条与改写的一条在修前代码上均失败（已实测：4 fail），修后通过。
|
||||
|
||||
## 门禁（Node 22.14）
|
||||
|
||||
| 项 | 基线 `f1a169cb` | 分支 |
|
||||
| --- | --- | --- |
|
||||
| tsc | — | 0 |
|
||||
| lint | — | 0 error / 126 warning |
|
||||
| npm test | 4,851 条，fail 24 | 4,854 条，fail 24（同名） |
|
||||
| build `/` | — | Static |
|
||||
| `.next/static` js gzip 合计 | 1,632,680 B | 1,632,662 B（−18 B） |
|
||||
|
||||
## 环境缺口
|
||||
|
||||
- 浏览器级（真机看进度栏）留给 `docs/testing/consult-plain-answer-20261001.md` 第 10 步。
|
||||
@@ -0,0 +1,49 @@
|
||||
# PROGRESS · 生时校正：健康区分题写入失败 + 按盘型挑题默认开启（2026-10-01）
|
||||
|
||||
执行：Claude 直接执行（产品负责人 2026-10-01「同意」：前后 10 分钟先开、修复后重测前后 30 分钟，过红线再放开）。分支 `codex/rectification-segment-order-20261001`。无单独任务书；背景见 `TASK-rectification-varga-resolution-20260930.md` T5 与 `docs/research/rectification_varga_resolution_2026_09_30.md` §4。BUG-1143、BUG-1144。
|
||||
|
||||
## 经过
|
||||
|
||||
1. 77 例 × ±10 / ±30 真库持久回放（PostgreSQL 17 替身；`frontend/tests/research/rectification-segment-persisted-replay.ts` 经 `scripts/research/varga_resolution_persisted_replay.py`）比较按段选题关 / 开：±10 开启头段多命中 6 例、真值段 0 丢失；±30 开启丢 3 例真值段（`docs/research/varga_resolution_persisted_onoff_before_fix_2026_10_01.json`）。
|
||||
2. 逐例插桩（临时，未提交）：丢失不是排序删题。区分类健康题写焦点时 `target_domain=health_pressure` 违反表约束，写入失败被吞，流程从 exhaustion 出口提前交付；开启排序只是把健康题排到第一位、更早撞上。默认顺序下同样发生（BUG-1143，BUG-672 同族复发）。
|
||||
3. 修 BUG-1143 后重跑（`…after_fix_2026_10_01.json`）。
|
||||
|
||||
## 77 例真库持久回放（修前 → 修后）
|
||||
|
||||
| 窗口 / 顺序 | 平均问题数 | 问满 6 题 | D9 头段命中 | D10 头段命中 | 真值段保留 D9 / D10 |
|
||||
| --- | --- | --- | --- | --- | --- |
|
||||
| ±10 默认 | 2.82 → 4.56 | 21 → 50 | 65 → 70 | 64 → 65 | 76/76 → 76/76 |
|
||||
| ±10 按段 | 2.87 → 4.61 | 22 → 52 | 68 → 71 | 67 → 67 | 76/76 → 76/76 |
|
||||
| ±30 默认 | 3.58 → 5.43 | 30 → 69 | 38 → 46 | 39 → 52 | 76/76 → 76/76 |
|
||||
| ±30 按段 | 3.53 → 5.43 | 29 → 69 | 41 → 51 | 40 → 55 | 74/75 → 76/76 |
|
||||
|
||||
修后日志 `retry next focus failed`：0 条。±60 未回放（排序包络上限 61 分钟，>61 不启用）。
|
||||
|
||||
## 改动
|
||||
|
||||
| 文件 | 改动 |
|
||||
| --- | --- |
|
||||
| `v9/server-focus.ts` | 新增 `focusTargetDomain`，所有意图焦点写入经 `persistableFocusDomain` |
|
||||
| `v9/method-followup.ts` | 三处焦点 ↔ 探针领域比较改 `sameCollectDomain` |
|
||||
| `core/segment-probe-order.ts` | `segmentOrderEnabledFor(window, flag)`：默认 ≤ 61 分钟开、`off` 关、`on` 研究包络 |
|
||||
| `v9/score-persist.ts` | 用上面函数代替 `RECTIFICATION_SEGMENT_ORDER === "on"` |
|
||||
| 测试 | `rectification-focus-target-domain-20261001.test.ts`（3）、`rectification-segment-order-default-20261001.test.ts`（2） |
|
||||
|
||||
既有断言改动：无。
|
||||
|
||||
## 门禁（对 `origin/staging`，替身真库除外均为本机）
|
||||
|
||||
| 检查 | 基线 | 本分支 |
|
||||
| --- | --- | --- |
|
||||
| tsc / lint | 0 / 0 error、126 warning | 0 / 0 error、126 warning |
|
||||
| npm test | 4846 / fail 24 | 4851 / fail 24；失败清单逐名相同，基线名 0 丢失，新增 5 |
|
||||
| 按镜像 COPY 清单构建 | — | exit 0 |
|
||||
| `/` / 首屏 gzip | Static / JS 555,622、CSS 68,813 | Static / 相同 |
|
||||
| 快速门 | — | Python 1036 passed / 1 skipped / 0 failed |
|
||||
|
||||
默认窗口由 21 放宽到 61 后：受影响 6 个测试文件 322 / 322、tsc 0。
|
||||
|
||||
## 缺口
|
||||
|
||||
- 真库证据为 PostgreSQL 17 替身；本次不改数据库结构。
|
||||
- 真机：部署后在 staging 走一次前后 30 分钟窗口的校正，确认能问到健康题、能问满几题。
|
||||
@@ -264,6 +264,7 @@
|
||||
| `TASK-rectification-message-cleanup-20261001.md` | `PROGRESS-rectification-message-cleanup-20261001.md` | **校正消息清理**：题干两遍（作废旧焦点仍打印题干 + 独立问题块再打印一次）→ 一屏一题干；证据轮正文改为「记下了 N 件事」+ 服务端清单可展开；校正结算后不显示「已完成 N 步」与「1.」泄漏。推翻 BUG-917 作废题留题干、VOICE 逐条复述 | 已验收（Claude 10-01：全套门禁对 staging 209352eb 一致；「1.」来源为推断——收起的有序列表复制补号，结算后整块不渲染已消除；真机待产品） | BUG-1135～1137 |
|
||||
| `TASK-rectification-chart-tier-caveats-20261001.md` | — | **报告与聊天标注判断不了的分盘**：采用后读 `active_birth_provenance`（segment-v1）→ 结果 `segment_summary` 档位，blocked/indistinct 的 D9/D10 在聊天 answer_policy 与一句说明、报告章节顶部提示 + 主题降级；旧资料逐字节不变；不改库、不改 consult/route.ts | 待领取 | BUG-1138～1139 |
|
||||
| `TASK-rectification-angle-timing-research-20261001.md` | `PROGRESS-rectification-angle-timing-research-20261001.md` | **角度类时间技法研究（离线）**：行运土木罗压本命上升 / 天顶度数、次限推运角、太阳弧角（1° ≈ 4 分钟出生时间），77 例 v5；乱序日期对照 ≥ 50 次必做、留一法、不得挤出真值；显著才叠加到六题后验比头段命中与区间宽度 | 已实现待验收（0/9 预登记检验显著，BUG-1141 closed_by_design；分支 codex/rectification-angle-timing-research-20261001，未推送） | BUG-1141 |
|
||||
| （直接执行，无任务书） | `PROGRESS-rectification-segment-order-20261001.md` | **健康区分题写入失败（BUG-1143）+ 按盘型挑题默认开启（BUG-1144）**：区分类焦点把 `health_pressure` 原样写入、违反表约束、写入被吞后提前交付（BUG-672 同族）；修后 77 例真库回放平均问题数 ±10 2.8→4.6、±30 3.6→5.4，D10 头段 ±30 39→52；按段选题 ≤61 分钟默认开，±30 D9 46→51、D10 52→55，真值段 76/76 | 已验收（Claude 10-01 直接执行） | 分支 codex/rectification-segment-order-20261001 |
|
||||
| `TASK-rectification-typed-event-scoring-research-20260929.md` | `PROGRESS-rectification-typed-event-research-20260929.md` | **打字经历按选择题规则计分（离线研究，不上线）**：计分通道不对称 + 已入账年份挡题；R0 学业质量题措辞 / 年精度显示成 1 月（冻结文件，需重新冻结） | 已验收关单（Claude 10-01：三类技法 9 项检验 0 显著，真值平均排位 0.49～0.59；Claude 补正对照——埋入信号排位 0.077、挪年 0.502——流程有效；BUG-1141 closed_by_design） | BUG-1088、1089 |
|
||||
| `TASK-report-reader-polish-20260929.md` | `PROGRESS-report-reader-polish-20260929.md` | **报告页打磨**：生成入口挪进页面主体（删标题栏按钮）、详情页到底部按钮（懒渲染一次到底）、导出按钮带文字、目录一级/二级分层、去掉「字段」与 RL/NL 缩写表头、状态列同义重复去重、报告表格淡底色、分块导出显示真实文件大小 | Claude 直接执行并自验（tsc 0、lint 0 error、全量 fail 与基线同 24 条、`/` Static、gzip +7 B、快速门 Python 1000 passed），已部署 staging `d7772011`，真机欠 | BUG-1092～1094 |
|
||||
| `TASK-report-english-edition-20260929.md` | `PROGRESS-report-english-edition-20260929.md` | **报告中英两版**：同一次引擎计算渲染 zh/en 两遍（不用模型翻译），瑜伽库 477 条补英文、模板与前端表头英文化；中文逐字节不变、英文零汉字、两版数字序列一致；阅读页 `?lang=en` 切换，导出当前语言 + 「问 AI 建议导出英文版」提示；旧报告不补英文 | Claude 直接执行（含两个 fork 子代理）并自验（tsc 0、lint 0 error、全量 fail 与基线同 24 条、`/` Static、gzip +0.06%、Python 定向全绿），已部署 staging `d7772011`，真机欠 | — |
|
||||
@@ -399,3 +400,4 @@
|
||||
- **实现合入 `staging` 的同一次推送里，必须同时把状态板那一行改掉。** 2026-09-16 的对账发现 10 份早已合入的单仍写着「待领取 / 待验收」，会导致重复派活。
|
||||
| `TASK-consult-plain-answer-20261001.md` | `PROGRESS-consult-plain-answer-20261001.md` | 普通对话「还是废话」：09-17 四步开场形状（格局名 → 谁推谁修 → 扮演哪个象）逼出谜语，问父母时爸妈被揉成一段；产品授权推翻该形状，改成先答 + 按问题里的对象分段 + 人话自检 + 空宫不单独下结论 + 父母卡标 mother/father（BUG-1132～1134） | **已实现，待 Claude 验收**（fork 子代理直接执行）；未推 staging、未部署；模型对比为环境缺口，真机清单 `docs/testing/consult-plain-answer-20261001.md` | 分支 `codex/consult-plain-answer-20261001` |
|
||||
| `TASK-consult-answer-clock-20261001.md` | `PROGRESS-consult-answer-clock-20261001.md` | 普通对话答题时钟 70 s 容不下推理模型（v4-pro 77 s），无出生分钟路线只受 110 s 工具钟管；产品定 5 分钟防卡死（BUG-1142） | **已实现，待 Claude 验收**（直接执行） | 分支 `codex/consult-answer-clock-20261001` |
|
||||
| `TASK-consult-timeline-no-plan-20261001.md` | `PROGRESS-consult-timeline-no-plan-20261001.md` | 普通对话进度栏提前打勾：回答提纲在算盘前就作为四行步骤显示并全部打勾（BUG-1145）；产品选 B 去掉提纲行，只显示真实步骤 | **实现完成，待 Claude 验收**（直接执行） | 分支 `codex/consult-timeline-no-plan-20261001` |
|
||||
|
||||
@@ -0,0 +1,31 @@
|
||||
# TASK · 普通对话进度栏去掉回答提纲行（2026-10-01）
|
||||
|
||||
- 基线：`origin/staging` `f1a169cb`
|
||||
- 分支 / worktree：`codex/consult-timeline-no-plan-20261001` / `.worktrees/consult-timeline-no-plan-20261001`
|
||||
- 模式：直接执行（产品 10-01 选 B 后明确要求直接做），Claude 派子代理实现、Claude 独立验收
|
||||
- BUG：BUG-1145
|
||||
|
||||
## 事故实证
|
||||
|
||||
产品 10-01 真机截图：问「我跟我父母的关系怎么样」，进度栏「读取分析方法」之后立刻出现「先回答你问的这件事 / 把判断钉在盘上 / 时间窗口怎么看 / 这周可以做什么」四行且全部打勾，最下面「正在计算本命盘…」仍在转。
|
||||
|
||||
## 根因
|
||||
|
||||
`frontend/src/lib/stream-agent-response.ts` `flushThinkingPlan` 在第一个流块就下发回答提纲（`think.plan` + 每节 `thinking.section`）；`frontend/src/lib/consultation-run-timeline.ts` `reduceConsultationTimeline` 把每节做成 live think 行，`completeLiveThink` 在下一个事件到来时把它们全标完成。提纲不是走过的步骤，却按步骤渲染。
|
||||
|
||||
## 决策记录
|
||||
|
||||
产品 10-01 在两案中选 **B：去掉提纲行，只显示真实步骤**（读取分析方法 → 计算本命盘 → 正在写 → 完成）。未选 A（提纲等算完再出现、按正文标题逐条打勾），原因：BUG-1132 之后回答按人分段、两节可省，提纲与正文对不上；产品偏好「多余入口宁可删除」。
|
||||
|
||||
## 硬红线
|
||||
|
||||
1. 只改渲染层（timeline reducer）；服务端事件、`thinkingPlan`、`applyThinkingSectionProgress`、存档字段、API / DB 结构不动。
|
||||
2. 生时校正进度栏不动（它不用这个 reducer，只共用行类型与组件）。
|
||||
3. `page.tsx` 不增长；不加 spinner / 骨架。
|
||||
|
||||
## 任务与验收
|
||||
|
||||
- T1 reducer：`think.plan` 不改状态；`thinking.section` 只暂存计算行明细并在计算行出现时补上；正文开始时计算行收口。验收：新增回归测试在修前失败、修后通过。
|
||||
- T2 历史重建与实时同形（`consultationTimelineFromSettled` 走同一 reducer；思考文本改走 `thinking.delta` 行）。验收：历史重建测试。
|
||||
- T3 记录：BUG_HISTORY、CHANGELOG、DESIGN.md、真机清单第 10 步、本任务书与进度记录、索引。
|
||||
- 门禁：tsc 0；lint 0 error；`npm test` 失败名单与基线同名；`next build` `/` Static；首屏 gzip ±2%。
|
||||
@@ -13,3 +13,4 @@
|
||||
| 7 | 第 1 步的回答如果明显被截断（最后一节没写完就停了） | 记下时间和模型名报给 Claude——这是取消字数上限后要盯的风险（见进度记录「长度副作用核对」） |
|
||||
| 8 | （BUG-1142，答题时钟改 5 分钟后）选 `deepseek-v4-pro` 新开对话，问「我和父母关系如何，他们怎么对待我」 | 回答完整写到「这周可以做的一件事」，不出现「回答未完成」；记下从发出到写完大约多少秒 |
|
||||
| 9 | 用没填出生分钟的人物档案、选 `deepseek-v4-pro` 问一个长问题（例如「完整讲讲我今年的事业和感情」） | 回答写完，不出现「回答未完成」 |
|
||||
| 10 | （BUG-1145）新开对话问任一问题，盯住回答上方的进度栏 | 只出现「读取分析方法」「计算本命盘」「正在写…」这类真实步骤；不再出现「先回答你问的这件事 / 把判断钉在盘上 / 时间窗口怎么看 / 这周可以做什么」四行，也没有在算盘时就打上的勾；刷新后打开这条历史对话，进度栏一样 |
|
||||
|
||||
+1
-1
@@ -482,7 +482,7 @@ The birth-time rectification session is the consultation transcript plus a house
|
||||
|
||||
#### Streaming states
|
||||
|
||||
Every assistant reply moves through the same states on both chat surfaces, and each state has exactly one visual. The step timeline (`ConsultationRunTimeline`) is the only activity surface. Consultation thinking is complete `think.step` items (v1 `thinking.delta` still renders for one release). Rectification never shows thinking text: provider reasoning is dropped at the public boundary.
|
||||
Every assistant reply moves through the same states on both chat surfaces, and each state has exactly one visual. The step timeline (`ConsultationRunTimeline`) is the only activity surface. Its rows are only steps the run has actually taken — reading the method, calculating the chart, writing (one row per heading the answer has written), and model reasoning when present. The answer outline the server plans (`think.plan` / `thinking.section`) is never a row: it only details the calculate row (2026-10-01, BUG-1145; it used to show four outline rows that ticked done the moment the calculation started). Consultation thinking is complete `think.step` items (v1 `thinking.delta` still renders for one release). Rectification never shows thinking text: provider reasoning is dropped at the public boundary.
|
||||
|
||||
| State | When | Visible | Transition in |
|
||||
|---|---|---|---|
|
||||
|
||||
@@ -30,6 +30,12 @@ export type ConsultationTimelineRow = Readonly<{
|
||||
export type ConsultationTimelineState = Readonly<{
|
||||
rows: readonly ConsultationTimelineRow[];
|
||||
answer: string;
|
||||
/**
|
||||
* What the run's planned sections say the calculation looks at, held until
|
||||
* the calculate row exists: the plan is sent before the tool starts, and it
|
||||
* only ever details that row (BUG-1145).
|
||||
*/
|
||||
calculateHints?: Readonly<{ queries: readonly string[]; sources: readonly string[] }>;
|
||||
}>;
|
||||
|
||||
const METHOD_ID = "method";
|
||||
@@ -55,12 +61,13 @@ export function reduceConsultationTimeline(
|
||||
return completeRow(state, METHOD_ID, CONSULTATION_DONE_SKILL_LABEL);
|
||||
}
|
||||
if (event.type === "tool.started") {
|
||||
return upsertRow(completeLiveThink(state), {
|
||||
const created = upsertRow(completeLiveThink(state), {
|
||||
id: CALCULATE_ID,
|
||||
kind: "calculate",
|
||||
status: "live",
|
||||
label: event.label || CONSULTATION_CHART_CALCULATION_LABEL,
|
||||
});
|
||||
return applyCalculateHints(created);
|
||||
}
|
||||
if (event.type === "activity") {
|
||||
if (event.phase === "chart-calculation") {
|
||||
@@ -88,18 +95,12 @@ export function reduceConsultationTimeline(
|
||||
if (event.type === "tool.failed") {
|
||||
return completeRow(state, CALCULATE_ID, CONSULTATION_DONE_CHART_LABEL);
|
||||
}
|
||||
if (event.type === "think.plan") {
|
||||
let next = completeLiveThink(state);
|
||||
for (const step of event.steps) {
|
||||
next = upsertRow(next, {
|
||||
id: `think-${step.id}`,
|
||||
kind: "think",
|
||||
status: "live",
|
||||
label: consultationThinkTitle(step.title),
|
||||
});
|
||||
}
|
||||
return next;
|
||||
}
|
||||
// The answer outline (think.plan / thinking.section) is what the answer
|
||||
// will be written as, not a step the run has taken, so it never becomes a
|
||||
// row: it was shown up front and ticked done as soon as the chart calculation
|
||||
// started, before a word of the answer existed (BUG-1145). The headings the
|
||||
// answer actually writes show as write rows below.
|
||||
if (event.type === "think.plan") return state;
|
||||
if (event.type === "think.step") {
|
||||
const id = `think-${event.id}`;
|
||||
const existing = state.rows.find((row) => row.id === id);
|
||||
@@ -127,14 +128,7 @@ export function reduceConsultationTimeline(
|
||||
heading: event.heading,
|
||||
steps: event.steps,
|
||||
};
|
||||
const withCalc = enrichCalculate(completeRow(state, CALCULATE_ID, CONSULTATION_DONE_CHART_LABEL), section);
|
||||
const withoutOpen = completeLiveThink(withCalc);
|
||||
return upsertRow(withoutOpen, {
|
||||
id: `think-${section.id}`,
|
||||
kind: "think",
|
||||
status: "live",
|
||||
label: consultationThinkTitle(section.title),
|
||||
});
|
||||
return applyCalculateHints(addCalculateHints(state, section));
|
||||
}
|
||||
if (event.type === "thinking.delta") {
|
||||
const think = lastRow(state.rows, (row) => row.kind === "think" && row.status === "live");
|
||||
@@ -152,7 +146,11 @@ export function reduceConsultationTimeline(
|
||||
});
|
||||
}
|
||||
if (event.type === "answer.delta") {
|
||||
const next = { ...completeLiveThink(state), answer: event.replace ? event.text : `${state.answer}${event.text}` };
|
||||
// Answer text only comes after the calculation, so the calculate row is
|
||||
// done by now even if its own completion event has not been seen (BUG-1145:
|
||||
// the planned sections used to close it).
|
||||
const calculated = completeRow(completeLiveThink(state), CALCULATE_ID, CONSULTATION_DONE_CHART_LABEL);
|
||||
const next = { ...calculated, answer: event.replace ? event.text : `${state.answer}${event.text}` };
|
||||
return syncWriteRows(next, false);
|
||||
}
|
||||
if (event.type === "run.completed" || event.type === "run.failed") {
|
||||
@@ -209,7 +207,7 @@ export function consultationTimelineFromSettled(input: {
|
||||
for (const section of sections) {
|
||||
events.push({ type: "thinking.section", ...section });
|
||||
}
|
||||
if (input.thinkingText?.trim() && sections.length === 0) {
|
||||
if (input.thinkingText?.trim()) {
|
||||
events.push({ type: "thinking.delta", text: input.thinkingText });
|
||||
}
|
||||
if (text) events.push({ type: "answer.delta", text });
|
||||
@@ -228,14 +226,7 @@ export function consultationTimelineFromSettled(input: {
|
||||
workflow: { route: "settled", status: "ready", preciseTiming: "blocked", missingLayers: [] },
|
||||
},
|
||||
});
|
||||
let state = reduceConsultationTimelineEvents(events);
|
||||
if (input.thinkingText?.trim() && sections.length > 0) {
|
||||
const lastThink = [...state.rows].reverse().find((row) => row.kind === "think");
|
||||
if (lastThink) {
|
||||
state = upsertRow(state, { ...lastThink, thinkingText: input.thinkingText.slice(0, 4_000) });
|
||||
}
|
||||
}
|
||||
return state.rows;
|
||||
return reduceConsultationTimelineEvents(events).rows;
|
||||
}
|
||||
|
||||
export function consultationThinkTitle(title: string): string {
|
||||
@@ -390,13 +381,25 @@ function syncWriteRows(state: ConsultationTimelineState, settled: boolean): Cons
|
||||
return next;
|
||||
}
|
||||
|
||||
function enrichCalculate(
|
||||
function addCalculateHints(
|
||||
state: ConsultationTimelineState,
|
||||
section: PublicThinkingSection,
|
||||
): ConsultationTimelineState {
|
||||
const hints = state.calculateHints;
|
||||
return {
|
||||
...state,
|
||||
calculateHints: {
|
||||
queries: uniqueLabels([...(hints?.queries ?? []), ...queriesFromSection(section)]).slice(0, 8),
|
||||
sources: uniqueLabels([...(hints?.sources ?? []), ...sourcesFromSection(section)]).slice(0, 8),
|
||||
},
|
||||
};
|
||||
}
|
||||
|
||||
function applyCalculateHints(state: ConsultationTimelineState): ConsultationTimelineState {
|
||||
const calculate = state.rows.find((row) => row.id === CALCULATE_ID);
|
||||
if (!calculate) return state;
|
||||
const queries = uniqueLabels([...(calculate.queries ?? []), ...queriesFromSection(section)]).slice(0, 8);
|
||||
const sources = uniqueLabels([...(calculate.sources ?? []), ...sourcesFromSection(section)]).slice(0, 8);
|
||||
const hints = state.calculateHints;
|
||||
if (!calculate || !hints) return state;
|
||||
const queries = uniqueLabels([...(calculate.queries ?? []), ...hints.queries]).slice(0, 8);
|
||||
const sources = uniqueLabels([...(calculate.sources ?? []), ...hints.sources]).slice(0, 8);
|
||||
return upsertRow(state, { ...calculate, queries, sources });
|
||||
}
|
||||
|
||||
@@ -62,3 +62,19 @@ export function orderProbesBySegment(input: {
|
||||
return input.probes.map((probe, index) => ({ probe, index, gain: segmentInformationGain(probe, weights, input.segments, input.offsets) }))
|
||||
.sort((a, b) => b.gain - a.gain || a.index - b.index).map((row) => row.probe);
|
||||
}
|
||||
|
||||
/**
|
||||
* Production default for segment-gain probe order (BUG-1144). 77-case persisted replay
|
||||
* (2026-10-01, after the health-focus write fix BUG-1143): at 21 and 61 minutes it gained
|
||||
* D9/D10 head hits with every truth segment kept (76/76). Before that fix the 61-minute cell lost
|
||||
* truth segments, which is why the first cut stopped at 21. Wider windows were not replayed.
|
||||
* "off" disables; "on" is the research envelope used by the persisted replay harness.
|
||||
*/
|
||||
export const SEGMENT_ORDER_DEFAULT_MAX_WINDOW = 61;
|
||||
export const SEGMENT_ORDER_RESEARCH_MAX_WINDOW = 61;
|
||||
|
||||
export function segmentOrderEnabledFor(windowMinutes: number | null | undefined, flag: string | undefined): boolean {
|
||||
if (flag === "off" || !windowMinutes || windowMinutes <= 0) return false;
|
||||
const limit = flag === "on" ? SEGMENT_ORDER_RESEARCH_MAX_WINDOW : SEGMENT_ORDER_DEFAULT_MAX_WINDOW;
|
||||
return windowMinutes <= limit;
|
||||
}
|
||||
|
||||
@@ -821,14 +821,14 @@ function liveDistinguishProbe(
|
||||
isValidDistinguishProbe({ ...probe, role: "distinguish" })
|
||||
&& (schemaKey
|
||||
? probe.semantic_key === schemaKey
|
||||
: (!focus.targetDomain || probe.domain === focus.targetDomain))
|
||||
: (!focus.targetDomain || sameCollectDomain(probe.domain, focus.targetDomain)))
|
||||
);
|
||||
const fromEvents = (eventProbes ?? []).find(matchEvent) ?? null;
|
||||
if (fromEvents) return fromEvents;
|
||||
const matchContrast = (probe: CandidateDiscriminatorProbe) => (
|
||||
schemaKey
|
||||
? probe.semanticKey === schemaKey
|
||||
: (!focus.targetDomain || probe.domain === focus.targetDomain)
|
||||
: (!focus.targetDomain || sameCollectDomain(probe.domain, focus.targetDomain))
|
||||
);
|
||||
const fromContrast = (contrastProbes ?? []).find(matchContrast);
|
||||
if (!fromContrast) return null;
|
||||
@@ -2580,7 +2580,7 @@ export function buildMethodFollowupPlan(input: {
|
||||
: themeFromDomain ?? (focus.intent === "out_of_sample_check" ? "oos_blind" : "education_style")
|
||||
) as MethodFollowup["ask_theme"];
|
||||
const matchingVerify = (input.eventProbes ?? []).find((probe) => (
|
||||
probe.domain === focus.targetDomain
|
||||
sameCollectDomain(probe.domain, focus.targetDomain)
|
||||
|| REVERSE_VERIFY_THEME[probe.domain as keyof typeof REVERSE_VERIFY_THEME] === keepAskTheme
|
||||
));
|
||||
const schema = focus.expectedAnswerSchema;
|
||||
|
||||
@@ -5,6 +5,7 @@
|
||||
* write a fresh inference_state, but with an empty answer ledger.
|
||||
*/
|
||||
import { shouldForceMinuteAfterSubBlocks } from "./block-scan.ts";
|
||||
import { segmentOrderEnabledFor } from "../core/segment-probe-order.ts";
|
||||
import { scanCaseSegments } from "./segment-scan.ts";
|
||||
import { targetChartsForDomain } from "../core/segment-summary.ts";
|
||||
import { buildProductCaseInferenceState } from "./product-inference.ts";
|
||||
@@ -450,7 +451,7 @@ export async function scoreAndPersistCurrentEvidence(input: {
|
||||
: undefined;
|
||||
const inference = buildProductCaseInferenceState({
|
||||
segmentMinutes, segmentTargets, segmentScanComplete: Boolean(segmentMinutes),
|
||||
segmentOrderEnabled: process.env.RECTIFICATION_SEGMENT_ORDER === "on",
|
||||
segmentOrderEnabled: segmentOrderEnabledFor(segmentMinutes?.length, process.env.RECTIFICATION_SEGMENT_ORDER),
|
||||
range: candidateRange,
|
||||
candidates: score.candidates,
|
||||
evidence: scorable,
|
||||
|
||||
@@ -272,6 +272,20 @@ export function persistableFocusDomain(domain: string | null | undefined): strin
|
||||
return null;
|
||||
}
|
||||
|
||||
/**
|
||||
* The domain a focus row may store. Every intent goes through persistableFocusDomain:
|
||||
* the table only accepts education/career/relationship/relocation/finance/health/family/other,
|
||||
* while planner domains include health_pressure and occupation (BUG-672). A raw
|
||||
* health_pressure on a distinguish focus failed the check constraint, the write was
|
||||
* swallowed and the session ended early (BUG-1143).
|
||||
*/
|
||||
export function focusTargetDomain(followup: Pick<MethodFollowup, "intent" | "domain">): string | null {
|
||||
return persistableFocusDomain(followup.domain)
|
||||
?? (followup.intent === "collect_method_evidence"
|
||||
? persistableFocusDomain(collectQuestionDomain(followup.domain))
|
||||
: null);
|
||||
}
|
||||
|
||||
function clampFocusTargetKind(kind: string | null | undefined): string | null {
|
||||
const value = kind?.trim() || "";
|
||||
if (!value) return null;
|
||||
@@ -672,10 +686,7 @@ async function persistServerOwnedFocusCore(input: {
|
||||
questionId,
|
||||
intent: followup.intent,
|
||||
targetEvidenceId: followup.date_reliability_evidence_id ?? null,
|
||||
targetDomain: followup.intent === "collect_method_evidence"
|
||||
? persistableFocusDomain(followup.domain)
|
||||
?? persistableFocusDomain(collectQuestionDomain(followup.domain))
|
||||
: followup.domain,
|
||||
targetDomain: focusTargetDomain(followup),
|
||||
targetKind: followup.intent === "collect_method_evidence"
|
||||
? collectFocusTargetKind(followup)
|
||||
: null,
|
||||
|
||||
@@ -9,7 +9,7 @@ import {
|
||||
reduceConsultationTimelineEvents,
|
||||
} from "../src/lib/consultation-run-timeline.ts";
|
||||
|
||||
test("career and wealth events append method, calculate 1/2, think, then the first write", () => {
|
||||
test("career and wealth events append method, calculate 1/2, then the first write (no outline row)", () => {
|
||||
const [question] = natalConsultationThinkingPlan({ domains: ["career", "wealth"] });
|
||||
assert.ok(question);
|
||||
const afterProgress = reduceConsultationTimelineEvents([
|
||||
@@ -39,13 +39,15 @@ test("career and wealth events append method, calculate 1/2, think, then the fir
|
||||
{ type: "answer.delta", text: "外松内紧。\n## 盘里支持这个判断的地方\n事业宫被土星压着。\n" },
|
||||
], afterProgress);
|
||||
|
||||
assert.deepEqual(state.rows.map((row) => row.kind), ["method", "calculate", "think", "write"]);
|
||||
// 原值: 行种类 ["method", "calculate", "think", "write"]，第 3 行是计划节「先回答你问的这件事」且已勾
|
||||
// 新值: 行种类 ["method", "calculate", "write"]，计划节不成行；计算行由正文开始收口
|
||||
// 原因: TASK-consult-timeline-no-plan-20261001（BUG-1145）：回答提纲不是已走过的步骤，产品 10-01 选 B 去掉提纲行
|
||||
assert.deepEqual(state.rows.map((row) => row.kind), ["method", "calculate", "write"]);
|
||||
assert.equal(state.rows[1]?.status, "done");
|
||||
assert.equal(state.rows[2]?.label, "先回答你问的这件事");
|
||||
assert.equal(state.rows[2]?.status, "done");
|
||||
assert.equal(state.rows[3]?.id, "write-盘里支持这个判断的地方");
|
||||
assert.equal(state.rows[3]?.status, "live");
|
||||
assert.match(state.rows[3]?.label ?? "", /正在写盘里支持这个判断的地方/);
|
||||
assert.equal(state.rows.some((row) => row.label === "先回答你问的这件事"), false);
|
||||
assert.equal(state.rows[2]?.id, "write-盘里支持这个判断的地方");
|
||||
assert.equal(state.rows[2]?.status, "live");
|
||||
assert.match(state.rows[2]?.label ?? "", /正在写盘里支持这个判断的地方/);
|
||||
});
|
||||
|
||||
test("thinking after a tool starts a new think row instead of reopening the first", () => {
|
||||
@@ -107,3 +109,62 @@ test("settled natal messages hydrate a calculate row from domain sections", () =
|
||||
const calculate = rows.find((row) => row.kind === "calculate");
|
||||
assert.ok(calculate?.queries?.some((item) => item.includes("事业") || item.includes("D10") || item.includes("财富")));
|
||||
});
|
||||
|
||||
// BUG-1145 (TASK-consult-timeline-no-plan-20261001): the answer outline was
|
||||
// shown as four think rows and ticked done the moment the calculation started.
|
||||
test("the answer outline sent before the calculation adds no row and ticks nothing", () => {
|
||||
const plan = natalConsultationThinkingPlan({ domains: ["parents"] });
|
||||
const state = reduceConsultationTimelineEvents([
|
||||
{ type: "run.started", runId: "run-1", requestId: "req-1" },
|
||||
{ type: "skill.started", name: "jyotish-vedic-astrology" },
|
||||
{ type: "skill.completed", name: "jyotish-vedic-astrology" },
|
||||
{ type: "think.plan", steps: plan.map((section) => ({ id: section.id, title: section.title })) },
|
||||
...plan.map((section) => ({ type: "thinking.section" as const, ...section })),
|
||||
{ type: "tool.started", callId: "tool-1", tool: "run-jyotish-consultation", label: "正在计算本命盘…" },
|
||||
]);
|
||||
assert.deepEqual(state.rows.map((row) => row.kind), ["method", "calculate"]);
|
||||
assert.deepEqual(state.rows.filter((row) => row.status === "done").map((row) => row.kind), ["method"]);
|
||||
assert.equal(state.rows[1]?.status, "live");
|
||||
for (const section of plan) {
|
||||
assert.equal(state.rows.some((row) => row.label === section.title), false, section.title);
|
||||
}
|
||||
// The outline still details what the calculation looks at.
|
||||
assert.ok((state.rows[1]?.queries?.length ?? 0) + (state.rows[1]?.sources?.length ?? 0) > 0);
|
||||
});
|
||||
|
||||
test("a finished run leaves no outline row and nothing live", () => {
|
||||
const plan = natalConsultationThinkingPlan({ domains: ["parents"] });
|
||||
const state = reduceConsultationTimelineEvents([
|
||||
{ type: "skill.started", name: "jyotish-vedic-astrology" },
|
||||
{ type: "skill.completed", name: "jyotish-vedic-astrology" },
|
||||
{ type: "think.plan", steps: plan.map((section) => ({ id: section.id, title: section.title })) },
|
||||
...plan.map((section) => ({ type: "thinking.section" as const, ...section })),
|
||||
{ type: "tool.started", callId: "tool-1", tool: "run-jyotish-consultation", label: "正在计算本命盘…" },
|
||||
{ type: "answer.delta", text: "你和父母不疏远。\n## 你和妈妈\n她操心多。\n## 这周可以做的一件事\n- 先问她最近怎样。\n" },
|
||||
{
|
||||
type: "run.completed",
|
||||
receipt: {
|
||||
runId: "run-1",
|
||||
runtime: "mastra-agentic",
|
||||
skill: { name: "jyotish-vedic-astrology", loaded: true, referenceReads: 0, methodologySections: 0 },
|
||||
steps: [],
|
||||
workflow: { route: "parents", status: "ready", preciseTiming: "blocked", missingLayers: [] },
|
||||
},
|
||||
},
|
||||
]);
|
||||
assert.deepEqual(state.rows.map((row) => row.kind), ["method", "calculate", "write", "write"]);
|
||||
assert.equal(state.rows.every((row) => row.status === "done"), true);
|
||||
assert.equal(state.rows.some((row) => row.kind === "think"), false);
|
||||
});
|
||||
|
||||
test("a stored natal answer rebuilds the same rows as live: no outline rows", () => {
|
||||
const plan = natalConsultationThinkingPlan({ domains: ["parents"] });
|
||||
const rows = consultationTimelineFromSettled({
|
||||
text: "你和父母不疏远。\n## 你和妈妈\n她操心多。\n## 你和爸爸\n他话少。\n",
|
||||
thinkingSections: plan,
|
||||
});
|
||||
assert.deepEqual(rows.map((row) => row.kind), ["method", "calculate", "write", "write"]);
|
||||
assert.equal(rows.some((row) => row.kind === "think"), false);
|
||||
assert.equal(rows.every((row) => row.status === "done"), true);
|
||||
assert.ok((rows[1]?.queries?.length ?? 0) + (rows[1]?.sources?.length ?? 0) > 0);
|
||||
});
|
||||
|
||||
@@ -0,0 +1,43 @@
|
||||
import assert from "node:assert/strict";
|
||||
import { readFileSync } from "node:fs";
|
||||
import test from "node:test";
|
||||
import { sameCollectDomain } from "../src/lib/rectification-agentic/v9/domain-alias.ts";
|
||||
import { focusTargetDomain } from "../src/lib/rectification-agentic/v9/server-focus.ts";
|
||||
|
||||
// BUG-1143: a distinguish focus wrote the planner domain verbatim. health_pressure is not in the
|
||||
// focus table's target_domain check, the write failed inside a swallowed retry, and the session
|
||||
// went straight to delivery although tap cards remained (BUG-672 alias family).
|
||||
const migration = readFileSync(new URL("../supabase/migrations/20260814020000_rectification_v10_runtime.sql", import.meta.url), "utf8");
|
||||
const allowedBlock = /target_domain is null or target_domain in \(([^)]*)\)/.exec(migration);
|
||||
assert.ok(allowedBlock, "focus table target_domain check is readable");
|
||||
const allowed = new Set([...allowedBlock[1]!.matchAll(/'([a-z_]+)'/g)].map((match) => match[1]!));
|
||||
|
||||
// Every domain the planner, engine catalog or collect pool can put on a followup.
|
||||
const plannerDomains = [
|
||||
"education", "relocation", "relationship", "career", "finance", "health_pressure", "family",
|
||||
"appearance", "occupation", "health", "other", "horary", "unknown", "active_focus",
|
||||
];
|
||||
|
||||
test("every followup domain maps to a value the focus table accepts, for every intent", () => {
|
||||
for (const intent of ["distinguish_candidates", "collect_method_evidence", "reverse_verify", "out_of_sample_check"] as const) {
|
||||
for (const domain of plannerDomains) {
|
||||
const stored = focusTargetDomain({ intent, domain } as never);
|
||||
assert.ok(stored === null || allowed.has(stored), `${intent}/${domain} -> ${stored}`);
|
||||
}
|
||||
}
|
||||
});
|
||||
|
||||
test("a health distinguish focus stores health and still matches health_pressure probes", () => {
|
||||
assert.equal(focusTargetDomain({ intent: "distinguish_candidates", domain: "health_pressure" } as never), "health");
|
||||
assert.equal(focusTargetDomain({ intent: "distinguish_candidates", domain: "occupation" } as never), "other");
|
||||
assert.equal(focusTargetDomain({ intent: "distinguish_candidates", domain: "relationship" } as never), "relationship");
|
||||
assert.equal(sameCollectDomain("health_pressure", "health"), true);
|
||||
});
|
||||
|
||||
test("focus writes and focus-to-probe matching never use a raw planner domain", () => {
|
||||
const serverFocus = readFileSync(new URL("../src/lib/rectification-agentic/v9/server-focus.ts", import.meta.url), "utf8");
|
||||
assert.doesNotMatch(serverFocus, /:\s*followup\.domain,\s*\n/, "no raw followup.domain passed as targetDomain");
|
||||
assert.match(serverFocus, /targetDomain: focusTargetDomain\(followup\)/);
|
||||
const followup = readFileSync(new URL("../src/lib/rectification-agentic/v9/method-followup.ts", import.meta.url), "utf8");
|
||||
assert.doesNotMatch(followup, /probe\.domain === focus\.targetDomain/);
|
||||
});
|
||||
@@ -0,0 +1,21 @@
|
||||
import assert from "node:assert/strict";
|
||||
import test from "node:test";
|
||||
import { segmentOrderEnabledFor, SEGMENT_ORDER_DEFAULT_MAX_WINDOW } from "../src/lib/rectification-agentic/core/segment-probe-order.ts";
|
||||
|
||||
// BUG-1144: segment-gain order is on by default only where the 77-case persisted replay lost no truth segment.
|
||||
test("segment order is on by default for windows up to 61 minutes and off beyond", () => {
|
||||
assert.equal(SEGMENT_ORDER_DEFAULT_MAX_WINDOW, 61);
|
||||
assert.equal(segmentOrderEnabledFor(21, undefined), true);
|
||||
assert.equal(segmentOrderEnabledFor(11, ""), true);
|
||||
assert.equal(segmentOrderEnabledFor(61, undefined), true);
|
||||
assert.equal(segmentOrderEnabledFor(62, undefined), false);
|
||||
assert.equal(segmentOrderEnabledFor(121, undefined), false);
|
||||
});
|
||||
|
||||
test("off disables everywhere; on keeps the research envelope; no scan means off", () => {
|
||||
assert.equal(segmentOrderEnabledFor(21, "off"), false);
|
||||
assert.equal(segmentOrderEnabledFor(61, "on"), true);
|
||||
assert.equal(segmentOrderEnabledFor(121, "on"), false);
|
||||
assert.equal(segmentOrderEnabledFor(undefined, undefined), false);
|
||||
assert.equal(segmentOrderEnabledFor(0, "on"), false);
|
||||
});
|
||||
Reference in New Issue
Block a user