fix(report): recommend by measured skin age and align admin results

docs/10.09新版H5实施与验收.md

@@ -1,6 +1,6 @@
 # 10.09 新版 H5 实施与验收
 
-更新：2026-10-10。
+更新：2026-10-12。
 
 ## 按确认项实现
 
@@ -18,7 +18,7 @@
 
 ## 文档未定义部分的实施口径
 
-话术文档没有新增计分公式或阈值，因此原六分类/三维度评分、肌龄计算及精华推荐规则不变。旧分档「预警/改善/良好」按低/中/高顺序映射为新文案的「改善/良好/优秀」。肤色和水分拆分展示，但继续共享原 `tone` 分类的分档，**不是新增独立评分算法**。
+话术文档没有新增计分公式或阈值，因此原六分类/三维度评分、肌龄计算不变。2026-10-12 经用户确认，精华推荐另按[肤龄基准与边界调整](产品推荐规则-肤龄版.md)实施，替代原三重判据；原始肌龄和展示肌龄在后台同时展示，历史快照不重算。旧分档「预警/改善/良好」按低/中/高顺序映射为新文案的「改善/良好/优秀」，后台同步此口径。肤色和水分拆分展示，但继续共享原 `tone` 分类的分档，**不是新增独立评分算法**。
 
 痘痘话术按真实脓包/结节计数决定是否输出中重度皮损说明；没有计数不宣称「未检出」。当真实计数与高分模板冲突时，展示层选择可反映计数的文案，不改变原分数。
 
@@ -31,6 +31,7 @@
 ```bash
 node h5/js/skin-score.js
 node h5/test/regression.js
+node h5/test/recommendation.js
 node h5/test/accounts.js
 node h5/test/report-persistence.mjs
 node h5/test/wechat-browser.mjs

docs/产品推荐规则-肤龄版.md

@@ -0,0 +1,67 @@
+# 产品推荐规则：肤龄基准与边界调整
+
+状态：2026-10-12 用户已确认三个边界条件；已实现。规则版本：`2026-10-12-skin-age-v1`。
+
+## 已明确的要求
+
+- 基准只取 Face++ 返回的原始 `skin_age`，不取实际年龄或修正后的展示肤龄。
+- 基础分组：肤龄小于 35 岁为 25+（主场精华 I），35 至不足 50 岁为 35+（II），50 岁及以上为 50+（III）。24 岁等不足 25 岁的检测结果也归 I。
+- 肤龄小于 32 岁，固定推荐 I，即使轮廓或皱纹得分低，也不触发升级。
+- 删除原来的实际年龄组、六分类均分、最差分类三重判据。
+- 边界附近结合三维度评分调整：紧塑力为必要条件，另外一个达标维度可为透亮感或稳韧力；不能仅因为另外两个维度高而降档。
+- 37 岁肌龄、仅存在暗沉问题的案例，应允许推荐 I。
+- 每份新报告仍只推荐一款；评分公式、检测流程和已有报告快照不因本次规则调整而改写。
+- 原始肤龄缺失或无效时，不回退实际年龄推测产品；需要明确显示本次无法推荐。
+
+## 已确认的三个条件
+
+1. 只在 35±3 岁及 50±3 岁范围内调整，范围外只按肤龄。
+2. 50 岁边界与 35 岁同样处理：47～不足50 岁紧塑力低于 8 分升至 III，50～53 岁紧塑力至少 9 分且另一维度至少 9 分降至 II。
+3. 「7+最低档」为低于 8 分；「9+」为大于等于 9 分，恰好 8 分不升档，恰好 9 分可以满足降档条件。
+
+## 决策顺序
+
+年龄支持小数，区间严格按下表执行：
+
+| 原始肤龄 | 默认产品 | 可调整条件 |
+|---|---|---|
+| 小于 32 | I | 固定，不升档 |
+| 32～不足35 | I | 紧塑力低于8，升为II |
+| 35～38（含38） | II | 紧塑力至少9，且透亮感/稳韧力任一至少9，降为I |
+| 大于38～不足47 | II | 按肤龄，不调整 |
+| 47～不足50 | II | 紧塑力低于8，升为III |
+| 50～53（含53） | III | 紧塑力至少9，且透亮感/稳韧力任一至少9，降为II |
+| 大于53 | III | 按肤龄，不调整 |
+
+调整最多一档。没有足够的必要维度数据证明满足调整条件时，保持肤龄基础档；不把缺失值当成低分。
+
+比较分数沿用评分引擎的三维度原始 `S=7+3×P/100`，不先进行前端一位小数或 VM 两位小数舍入。例如 7.99999 小于 8，8.99999 小于 9。原六分类和三维度评分公式、低/中/高档位阈值均不变；推荐阈值与档位文案阈值是不同概念。
+
+## 后台与前端一致性
+
+- 后台列表新增「肌龄（原始 / 展示）」；档案里的年龄明确为实际年龄。
+- 名称统一为紧塑力、透亮感、稳韧力；列表显示与 H5 相同的一位小数分数。内部旧档位 `预警/改善/良好` 映射成 `改善/良好/优秀`，筛选参数仍传原值，避免改写旧数据或破坏统计。
+- 新记录同时保存 `skin_age_display`、`recommendation` 和前端 `presentation`。后台详情展示基础档、最终档、未舍入的判定分、推荐原因与规则版本；七项结论和护理建议直接使用同一份前端快照。
+- 旧记录没有新规则元数据时，明确显示「旧版规则（历史记录）」，不按新规则重新推荐。后台详情可读取原账户已保存的前端展示快照；无快照则标注旧版文案。旧记录未存展示肌龄时，按原有实际年龄±3规则只读补全展示值。
+- CSV 同步提供两种肌龄、统一的新档位名、推荐版本和原因。
+
+## 实施与验证范围
+
+- `h5/js/skin-score.js`：推荐决策、产品组别、推荐原因及相关综合文案；保留原评分算法。
+- `h5/js/report-ui.js`：前后台共用名称、档位、三款产品映射；H5 无有效肤龄时隐藏具体产品并说明原因。
+- 更新引擎自测和 Excel 回归中的产品相关断言；原始 Excel 保留为旧算法参照，不改写评分真值来掩盖差异。
+- 新增边界测试：24岁且轮廓/皱纹低分；31.99/32/34.99/35/38/38.01；46.99/47/49.99/50/53/53.01；紧塑力7.99/8/8.99/9；另两个维度分别达标；必要条件不满足；实际年龄改变不影响推荐；原始肤龄和显示肤龄不同；维度或肤龄缺失。
+- 验证新报告保存与重开后产品一致，已有报告仍读取原有快照。
+- 本次不调整展示肤龄的实际年龄±3岁修正规则，后台并列展示两者，推荐只使用原始肤龄。
+
+验证命令：
+
+```bash
+node h5/test/recommendation.js
+node h5/js/skin-score.js
+node h5/test/regression.js
+node h5/test/accounts.js
+node h5/test/report-persistence.mjs
+```
+
+已通过 126 个推荐用例、83 条评分引擎自测、86 条真值回归。隔离 HTTP + Chromium 验证前后台原始/展示肌龄、三维度评分、推荐产品、七项结论、护理建议、筛选与 CSV 一致；验证保存重开、旧报告快照不改写、无肤龄时不残留上一款产品。全部使用样本上游，不调用收费 Face++。

graphify-out/.graphify_labels.json

@@ -3,31 +3,31 @@
   "1": "线上闭环测试（不含真实微信授权）",
   "2": "app.js",
   "3": "skin-score.js",
-  "4": "n",
-  "5": "normalize",
-  "6": "getContext2dOrThrow",
-  "7": "extractFaces",
+  "4": "i",
+  "5": "n",
+  "6": ".runAndExtendWithFaceDetections",
+  "7": "Box",
   "8": "admin.js",
   "9": "__awaiter",
   "10": "AgeGenderNet",
   "11": "lib/accounts.js",
   "12": "server.js",
   "13": "files",
-  "14": "FaceLandmarks",
+  "14": "FaceLandmarks68",
   "15": ".runAndExtendWithFaceDetection",
   "16": "巨子测肤 · 1'me 在线肌肤测试 — UI 设计规范",
-  "17": "getEnv",
+  "17": "extractFaces",
   "18": "1'me 在线肌肤测试 H5",
   "19": "head-pose.js",
   "20": "build_fixtures.py",
   "21": "disposeUnusedWeightTensors",
-  "22": ".computeFaceDescriptor",
+  "22": "ComputeAllFaceDescriptorsTask",
   "23": "facepp-skinanalyze（旷视皮肤分析专业版调用技能）",
   "24": "巨子测肤 · 1'me 语音播报内容设计文档 V1.0",
   "25": "mobile-select.iife.min.js",
   "26": "wi",
   "27": "gen_voice.py",
-  "28": "wn",
+  "28": "C",
   "29": "face-api.js 本地化依赖（h5/vendor/faceapi/）",
   "30": "ui-20261009/manifest.json",
   "31": "browser-failures.mjs",
@@ -35,49 +35,51 @@
   "33": "build-fixture.py",
   "34": "smoke-test.js",
   "35": "DetectSingleFaceLandmarksTask",
-  "36": "so",
+  "36": "ao",
   "37": "UI 外发版 10.09 图片资源",
-  "38": "PredictSingleAgeAndGenderWithFaceAlignmentTask",
+  "38": "FaceLandmark68Net",
   "39": "UI交互界面调整",
-  "40": "extractAllFacesAndComputeResults",
-  "41": "initialize",
+  "40": ".predictAgeAndGender",
+  "41": "test/accounts.js",
   "42": "export_psd_assets.py",
-  "43": "FaceLandmark68Net",
+  "43": ".loadFromDisk",
   "44": "start-local.sh",
   "45": "PredictAllAgeAndGenderWithFaceAlignmentTask",
   "46": "PredictAllFaceExpressionsWithFaceAlignmentTask",
-  "47": "TinyYolov2Base",
-  "48": "PredictSingleFaceExpressionsWithFaceAlignmentTask",
+  "47": "recommendation.js",
+  "48": "report-persistence.mjs",
   "49": "mobile-select/README.md",
   "50": "call_skinanalyze_3view.sh",
   "51": "第二轮2反馈.md",
   "52": "region-data.js",
   "53": "护理建议（全量文本，按需求摘抄）",
   "54": "UI 对齐 10.09 设计稿：调研与实施计划",
-  "55": "NeuralNetwork",
-  "56": "convLayer",
+  "55": "extractFaceTensors",
+  "56": "report-ui.js",
   "57": "regression.js",
   "58": ".dispose",
   "59": "call_skinanalyze.sh",
   "60": "report-fixture.js",
-  "61": ".extractWeights",
+  "61": "NeuralNetwork",
   "62": "提交前 Review（2026-10-11）",
   "63": "e2e-browser.mjs",
   "64": "account-client.js",
-  "65": "tt",
-  "66": "cn",
+  "65": "k",
+  "66": "wechat-browser.mjs",
   "67": "ui-20261009.mjs",
   "68": "analyze-timeout.js",
   "69": "misans/README.md",
-  "70": "decodeBoxesLayer",
-  "71": "dr",
+  "70": ".draw",
+  "71": "production-loop.mjs",
   "72": "build_report_copy.py",
   "73": "camera-geometry.js",
   "74": "report-copy.js",
   "75": "prepare_h5_assets.py",
   "76": "TinyYolov2",
-  "77": "FaceDetection",
+  "77": "drawFaceExpressions",
+  "78": ".toString",
   "79": "isValidNumber",
   "80": "线上微信登录部署记录",
-  "83": "Yo"
+  "81": "ComputeSingleFaceDescriptorTask",
+  "82": "PredictAllFaceExpressionsTask"
 }

graphify-out/.graphify_labels.json.sig

@@ -1 +1 @@
-{"0": "69ee47981e97a7eb", "1": "cb6ea50bd4e4a5ca", "2": "9cf893b45c314643", "3": "8ac814cdb72a1a7d", "4": "0f24ec9aab01e512", "5": "0e8200cefc11b932", "6": "1169f97bb458af6f", "7": "c1ae3952d970e510", "8": "cf2aba493368e557", "9": "8abf7998360e0ccc", "10": "c796af251148a471", "11": "79c3e26b93d030ce", "12": "ab705784bbd1dbdc", "13": "7e3b558a587a4ebe", "14": "186e1f68d4b2dfe9", "15": "5a9014c4a7f87699", "16": "95add6a975c0cb72", "17": "773ca6b3d5102a78", "18": "5aa9c1fc3b3c8196", "19": "0a373ad2197fd1a6", "20": "aecf2fce44ba49c5", "21": "5d94cf57d5d1a6fb", "22": "444ef0744645f5ac", "23": "90fac00055070ce2", "24": "f85b5b1634224c86", "25": "46e8db3e6e3f37ea", "26": "83b0fd3b14363e4d", "27": "eab427ce9a5219e9", "28": "cf86030ab36386aa", "29": "c608f1c6492e1cb1", "30": "73c13793fca09f56", "31": "fbbcfc059751cc6a", "32": "f6bcf56c4589316b", "33": "75f750c5b7d9f63d", "34": "f024ff7942935690", "35": "23f1d406f6c779ae", "36": "c55f872b3e16ab6f", "37": "cb8f26249485324a", "38": "01f68dfc34a29d32", "39": "fdd95375b4a5ddb3", "40": "3a68564642f68c0d", "41": "eb3733e409dc7411", "42": "ffd960136fde00ad", "43": "9a25b1230b5fe649", "44": "852c1bb2446edd9c", "45": "92ab3e5134ed8685", "46": "65718282b3c32109", "47": "41f3bef447e57f23", "48": "8c471c7baeaaca54", "49": "1f976711cda158dc", "50": "bbb2b57d72922b16", "51": "eb6a182204867275", "52": "2ec3c8c088568ba9", "53": "68c45056378341a7", "54": "c58d41e8e580fdc6", "55": "d96493a5f110bd4c", "56": "e649ddeadda89cbb", "57": "f2a40b5f1b29f669", "58": "73a31170f76a338b", "59": "891f59cda3708718", "60": "a505b7de6257a4e8", "61": "400fa8ac8b726a66", "62": "1366ae24c7e5897d", "63": "c256b43f75d7809e", "64": "86101547c05f96b9", "65": "3088ae48043ee39d", "66": "932f010dc6115fb4", "67": "3e126d8deaf6e200", "68": "6b91f3679541de79", "69": "e511846eeba03076", "70": "c396c038e5f5007f", "71": "2bfc96d4090afd84", "72": "c84f196a7bd35861", "73": "2b887bf601ccfb9f", "74": "ae2a154311e1f53a", "75": "69fdd23278518884", "76": "29790cf710874dd0", "77": "287d60115d9acda8", "79": "8f647d3ce08ecd2b", "80": "059d6ba8ca7ec3a0", "83": "a54008a71fdd44fa"}
+{"0": "73743c18a4a7fca9", "1": "cb6ea50bd4e4a5ca", "2": "9cf893b45c314643", "3": "1f737ca8edb978f2", "4": "55697fad20675b78", "5": "be544c1e1f2ddb64", "6": "46204fb2e4d50c53", "7": "fcfa9f1898866dc8", "8": "1ecf47531e1d507b", "9": "2e6b02a61343f092", "10": "563202faef7b3421", "11": "edc9f95405704875", "12": "d3e946c5907aa577", "13": "7e3b558a587a4ebe", "14": "15606ce5d8eec23a", "15": "5a9014c4a7f87699", "16": "95add6a975c0cb72", "17": "2405e1edb003a83e", "18": "300b45cf926eb704", "19": "0a373ad2197fd1a6", "20": "aecf2fce44ba49c5", "21": "d3cd661c24264482", "22": "3dd68cce72af5314", "23": "90fac00055070ce2", "24": "f85b5b1634224c86", "25": "46e8db3e6e3f37ea", "26": "6b6c40a418e18aed", "27": "eab427ce9a5219e9", "28": "00f6d1f8841af9c4", "29": "c608f1c6492e1cb1", "30": "73c13793fca09f56", "31": "fbbcfc059751cc6a", "32": "f6bcf56c4589316b", "33": "75f750c5b7d9f63d", "34": "f024ff7942935690", "35": "23f1d406f6c779ae", "36": "8e5e97153c881130", "37": "cb8f26249485324a", "38": "e2d12f8289f83819", "39": "fdd95375b4a5ddb3", "40": "a2b13160223b5a98", "41": "4bc370be8afd7db1", "42": "ffd960136fde00ad", "43": "00dab8704da757dc", "44": "852c1bb2446edd9c", "45": "92ab3e5134ed8685", "46": "65718282b3c32109", "47": "e594ecaa8e7d1538", "48": "e042d7fe7351bcb9", "49": "1f976711cda158dc", "50": "bbb2b57d72922b16", "51": "eb6a182204867275", "52": "2ec3c8c088568ba9", "53": "68c45056378341a7", "54": "c58d41e8e580fdc6", "55": "0d376c2f52a725e0", "56": "91d7ab49459e0b92", "57": "f2a40b5f1b29f669", "58": "675690fb61ef5bc7", "59": "891f59cda3708718", "60": "a505b7de6257a4e8", "61": "18ffa2c1feddfd10", "62": "1366ae24c7e5897d", "63": "c256b43f75d7809e", "64": "86101547c05f96b9", "65": "b7874aef5ca3f2ee", "66": "2b44d0efd8eb25c5", "67": "3e126d8deaf6e200", "68": "6b91f3679541de79", "69": "e511846eeba03076", "70": "bf75bfc99d4f9e34", "71": "4195478d5c528371", "72": "c84f196a7bd35861", "73": "2b887bf601ccfb9f", "74": "ae2a154311e1f53a", "75": "69fdd23278518884", "76": "29790cf710874dd0", "77": "a0bc88eb14b70170", "78": "42c9bc548e269a96", "79": "1c27e5f448dc4f9c", "80": "059d6ba8ca7ec3a0", "81": "7ac475ea9bc280f8", "82": "e800d832c255899a"}

graphify-out/GRAPH_REPORT.md

@@ -1,16 +1,16 @@
-# Graph Report - juzi  (2026-10-11)
+# Graph Report - juzi  (2026-10-12)
 
 ## Corpus Check
-- 64 files · ~4,182,459 words
+- 66 files · ~4,184,136 words
 - Verdict: corpus is large enough that graph structure adds value.
 
 ## Summary
-- 1164 nodes · 2523 edges · 81 communities (63 shown, 18 thin omitted)
+- 1188 nodes · 2566 edges · 83 communities (66 shown, 17 thin omitted)
 - Extraction: 93% EXTRACTED · 7% INFERRED · 0% AMBIGUOUS · INFERRED: 179 edges (avg confidence: 0.53)
 - Token cost: 0 input · 0 output
 
 ## Graph Freshness
-- Built from commit: `3dda4ec7`
+- Built from commit: `7518dc23`
 - Run `git rev-parse HEAD` and compare to check if the graph is stale.
 - Run `graphify update .` after code changes (no API cost).
 
@@ -19,76 +19,78 @@
 - 线上闭环测试（不含真实微信授权）
 - app.js
 - skin-score.js
+- i
 - n
-- normalize
-- getContext2dOrThrow
-- extractFaces
+- .runAndExtendWithFaceDetections
+- Box
 - admin.js
 - __awaiter
 - AgeGenderNet
 - lib/accounts.js
 - server.js
 - files
-- FaceLandmarks
+- FaceLandmarks68
 - .runAndExtendWithFaceDetection
 - 巨子测肤 · 1'me 在线肌肤测试 — UI 设计规范
-- getEnv
+- extractFaces
 - 1'me 在线肌肤测试 H5
 - head-pose.js
 - build_fixtures.py
 - disposeUnusedWeightTensors
-- .computeFaceDescriptor
+- ComputeAllFaceDescriptorsTask
 - facepp-skinanalyze（旷视皮肤分析专业版调用技能）
 - 巨子测肤 · 1'me 语音播报内容设计文档 V1.0
 - mobile-select.iife.min.js
 - wi
 - gen_voice.py
-- wn
+- C
 - face-api.js 本地化依赖（h5/vendor/faceapi/）
 - ui-20261009/manifest.json
 - browser-failures.mjs
 - FaceMatcher
 - build-fixture.py
 - DetectSingleFaceLandmarksTask
-- so
+- ao
 - UI 外发版 10.09 图片资源
-- PredictSingleAgeAndGenderWithFaceAlignmentTask
-- UI交互界面调整
-- extractAllFacesAndComputeResults
-- initialize
-- export_psd_assets.py
 - FaceLandmark68Net
+- UI交互界面调整
+- .predictAgeAndGender
+- test/accounts.js
+- export_psd_assets.py
+- .loadFromDisk
 - start-local.sh
 - PredictAllAgeAndGenderWithFaceAlignmentTask
 - PredictAllFaceExpressionsWithFaceAlignmentTask
-- TinyYolov2Base
-- PredictSingleFaceExpressionsWithFaceAlignmentTask
+- recommendation.js
+- report-persistence.mjs
 - mobile-select/README.md
 - call_skinanalyze_3view.sh
 - 护理建议（全量文本，按需求摘抄）
 - UI 对齐 10.09 设计稿：调研与实施计划
-- NeuralNetwork
-- convLayer
+- extractFaceTensors
+- report-ui.js
 - regression.js
 - .dispose
 - call_skinanalyze.sh
-- .extractWeights
+- NeuralNetwork
 - 提交前 Review（2026-10-11）
 - e2e-browser.mjs
 - account-client.js
-- tt
-- cn
+- k
+- wechat-browser.mjs
 - ui-20261009.mjs
 - analyze-timeout.js
 - misans/README.md
-- decodeBoxesLayer
-- dr
+- .draw
+- production-loop.mjs
 - build_report_copy.py
 - TinyYolov2
-- FaceDetection
+- drawFaceExpressions
+- .toString
 - isValidNumber
 - 线上微信登录部署记录
-- Yo
+- ComputeSingleFaceDescriptorTask
+- PredictAllFaceExpressionsTask
 
 ## God Nodes (most connected - your core abstractions)
 1. `__awaiter()` - 59 edges
@@ -100,7 +102,7 @@
 7. `C()` - 32 edges
 8. `toNetInput()` - 27 edges
 9. `e()` - 22 edges
-10. `f()` - 20 edges
+10. `loadRecords()` - 20 edges
 
 ## Surprising Connections (you probably didn't know these)
 - `loadRecords()` --indirect_call--> `it()`  [INFERRED]
@@ -117,11 +119,11 @@
 ## Import Cycles
 - None detected.
 
-## Communities (81 total, 18 thin omitted)
+## Communities (83 total, 17 thin omitted)
 
 ### Community 0 - "face-api.js"
 Cohesion: 0.02
-Nodes (57): bn(), Bo(), ca(), cd(), ce(), computeReshapedDimensions(), dn(), ee() (+49 more)
+Nodes (35): cn(), computeReshapedDimensions(), createBrowserEnv(), createFileSystem(), createNodejsEnv(), dr(), en(), FaceDetection() (+27 more)
 
 ### Community 1 - "线上闭环测试（不含真实微信授权）"
 Cohesion: 0.25
@@ -133,51 +135,47 @@ Nodes (74): applyMirror(), armNudge(), beginDetect(), bindBackdropClose(), busyW
 
 ### Community 3 - "skin-score.js"
 Cohesion: 0.12
-Nodes (44): acneGrade(), buildViewModel(), calcAcne(), calcContour(), calcSens(), calcSmooth(), calcTone(), calcWrinkles() (+36 more)
+Nodes (45): acneGrade(), buildViewModel(), calcAcne(), calcContour(), calcSens(), calcSmooth(), calcTone(), calcWrinkles() (+37 more)
 
-### Community 4 - "n"
-Cohesion: 0.07
-Nodes (88): a(), ae(), ao(), b(), ba(), be(), bf(), bi() (+80 more)
+### Community 4 - "i"
+Cohesion: 0.09
+Nodes (42): a(), bf(), cf(), d(), da(), dn(), DrawBox(), drawContour() (+34 more)
 
-### Community 5 - "normalize"
+### Community 5 - "n"
 Cohesion: 0.11
-Nodes (17): conv(), conv$1(), convDown(), convLayer$1(), convNoRelu(), denseBlock3(), denseBlock4(), depthwiseSeparableConv() (+9 more)
+Nodes (37): ae(), b(), be(), bi(), ce(), de(), e(), Ef() (+29 more)
 
-### Community 6 - "getContext2dOrThrow"
-Cohesion: 0.12
-Nodes (15): DetectAllFacesTask(), DrawBox(), drawDetections(), drawFaceExpressions(), DrawFaceLandmarks(), DrawTextField(), extendWithFaceDetection(), FaceExpressions() (+7 more)
-
-### Community 7 - "extractFaces"
-Cohesion: 0.15
-Nodes (5): Box(), DetectAllFaceLandmarksTask(), extractFaces(), extractFaceTensors(), isTensor3D()
+### Community 7 - "Box"
+Cohesion: 0.10
+Nodes (11): Box(), extendWithFaceDescriptor(), extendWithFaceDetection(), extendWithFaceLandmarks(), FaceLandmarks(), FaceRecognitionNet(), isWithFaceDetection(), isWithFaceLandmarks() (+3 more)
 
 ### Community 8 - "admin.js"
-Cohesion: 0.25
-Nodes (23): api(), bind(), closeDetail(), collectFilters(), dimsHtml(), doRetry(), dSec(), el() (+15 more)
+Cohesion: 0.21
+Nodes (28): ageText(), api(), bind(), closeDetail(), collectFilters(), dimsHtml(), doRetry(), dSec() (+20 more)
 
 ### Community 9 - "__awaiter"
 Cohesion: 0.14
-Nodes (12): __awaiter(), ComposableTask(), fetchImage(), fetchJson(), fetchNetWeights(), fetchOrThrow(), __generator(), getModelUris() (+4 more)
+Nodes (13): __awaiter(), extendWithFaceExpressions(), extractAllFacesAndComputeResults(), extractSingleFaceAndComputeResult(), fetchImage(), fetchJson(), fetchNetWeights(), fetchOrThrow() (+5 more)
 
 ### Community 10 - "AgeGenderNet"
-Cohesion: 0.11
-Nodes (5): AgeGenderNet(), FaceExpressionNet(), FaceLandmark68NetBase(), FaceLandmark68TinyNet(), FaceProcessor()
+Cohesion: 0.05
+Nodes (31): AgeGenderNet(), conv(), conv$1(), convDown(), convLayer(), convLayer$1(), convNoRelu(), convWithBatchNorm() (+23 more)
 
 ### Community 11 - "lib/accounts.js"
-Cohesion: 0.06
-Nodes (43): build(), count(), fill(), numeric(), atomic(), body(), cookie(), cookies() (+35 more)
+Cohesion: 0.17
+Nodes (19): atomic(), body(), cookie(), cookies(), crypto, fs, json(), login() (+11 more)
 
 ### Community 12 - "server.js"
 Cohesion: 0.08
-Nodes (48): accounts, appendRecord(), callFacepp(), CRED, csvCell(), DATA_DIR, dumpRaw(), ERR_MSG (+40 more)
+Nodes (49): accounts, appendRecord(), callFacepp(), CRED, csvCell(), DATA_DIR, displaySkinAge(), dumpRaw() (+41 more)
 
 ### Community 13 - "files"
 Cohesion: 0.12
 Nodes (15): files, analyze-fail, analyzing, countdown-321, hint-far, hint-front, hint-hold, hint-near (+7 more)
 
-### Community 14 - "FaceLandmarks"
-Cohesion: 0.10
-Nodes (9): bgrToRgbTensor(), extendWithFaceLandmarks(), FaceLandmarks(), FaceLandmarks5(), FaceLandmarks68(), getCenterPoint(), isDimensions(), minBbox() (+1 more)
+### Community 14 - "FaceLandmarks68"
+Cohesion: 0.19
+Nodes (3): FaceLandmarks5(), FaceLandmarks68(), getCenterPoint()
 
 ### Community 15 - ".runAndExtendWithFaceDetection"
 Cohesion: 0.33
@@ -187,13 +185,13 @@ Nodes (5): allFacesMtcnn(), allFacesSsdMobilenetv1(), allFacesTinyYolov2(), dete
 Cohesion: 0.15
 Nodes (12): 1. 基准与尺寸换算, 2. 色彩规范, 3. 字体规范, 4. 组件切图(assets/svg/), 5. 图片素材(assets/images/,共 15 张), 6. 页面清单(design/ 下有整屏 PNG 预览), 7. 目录结构, 8. 还原建议 (+4 more)
 
-### Community 17 - "getEnv"
+### Community 17 - "extractFaces"
 Cohesion: 0.19
-Nodes (14): awaitMediaLoaded(), bufferToImage(), createCanvas(), createCanvasFromMedia(), extractImagePatches(), getEnv(), getMediaDimensions(), imageTensorToCanvas() (+6 more)
+Nodes (19): awaitMediaLoaded(), bufferToImage(), createCanvas(), createCanvasFromMedia(), extractFaces(), extractImagePatches(), getContext2dOrThrow(), getEnv() (+11 more)
 
 ### Community 18 - "1'me 在线肌肤测试 H5"
-Cohesion: 0.07
-Nodes (27): 10.09 新版 H5 实施与验收, 按确认项实现, 文档未定义部分的实施口径, 素材再生成, 验证命令, 公网服务器配置, 公网真机验收清单, 反向代理与后台保护 (+19 more)
+Cohesion: 0.05
+Nodes (33): 10.09 新版 H5 实施与验收, 按确认项实现, 文档未定义部分的实施口径, 素材再生成, 验证命令, 产品推荐规则：肤龄基准与边界调整, 决策顺序, 后台与前端一致性 (+25 more)
 
 ### Community 19 - "head-pose.js"
 Cohesion: 0.33
@@ -204,12 +202,8 @@ Cohesion: 0.40
 Nodes (9): build_case1(), build_case2(), build_truth(), col(), dump(), main(), parse_value(), set_path() (+1 more)
 
 ### Community 21 - "disposeUnusedWeightTensors"
-Cohesion: 0.09
-Nodes (23): disposeUnusedWeightTensors(), extractorsFactory$3(), extractorsFactory$5(), extractorsFactory$7(), extractorsFactory$9(), extractParamsFromWeigthMap(), extractParamsFromWeigthMap$1(), extractParamsFromWeigthMap$2() (+15 more)
-
-### Community 22 - ".computeFaceDescriptor"
-Cohesion: 0.18
-Nodes (4): ComputeAllFaceDescriptorsTask(), ComputeSingleFaceDescriptorTask(), extendWithFaceDescriptor(), FaceRecognitionNet()
+Cohesion: 0.06
+Nodes (33): boxPredictionLayer(), decodeBoxesLayer(), depthwiseConvLayer(), disposeUnusedWeightTensors(), extractorsFactory$3(), extractorsFactory$5(), extractorsFactory$7(), extractorsFactory$9() (+25 more)
 
 ### Community 23 - "facepp-skinanalyze（旷视皮肤分析专业版调用技能）"
 Cohesion: 0.20
@@ -224,16 +218,16 @@ Cohesion: 0.20
 Nodes (16): _arrayLikeToArray(), _arrayWithHoles(), _arrayWithoutHoles(), _classCallCheck(), _createClass(), _defineProperties(), h(), _iterableToArray() (+8 more)
 
 ### Community 26 - "wi"
-Cohesion: 0.28
-Nodes (9): Ci(), Ei(), Fi(), Ii(), ki(), pe(), ri(), wi() (+1 more)
+Cohesion: 0.20
+Nodes (12): Ci(), Ei(), Fi(), Ii(), ki(), pe(), Qt(), ri() (+4 more)
 
 ### Community 27 - "gen_voice.py"
 Cohesion: 0.30
 Nodes (11): ndarray, Path, gen_countdown(), main(), normalize_clip(), 倒计时单独按字生成再拼接，避免「三、二、一」整句被读成播报腔、与提示音气质漂移。 与提示音共用同一 speaker + instruct；句间留约…, 裁掉过长首尾静音，保留 ≤150ms 边沿，避免提示显迟钝。, 响度归一 −16 LUFS + 转 MP3 48k CBR mono 22050Hz。 (+3 more)
 
-### Community 28 - "wn"
-Cohesion: 0.67
-Nodes (3): wl(), wn(), xn()
+### Community 28 - "C"
+Cohesion: 0.08
+Nodes (33): bn(), Bo(), C(), cd(), cl(), co(), eo(), fn() (+25 more)
 
 ### Community 29 - "face-api.js 本地化依赖（h5/vendor/faceapi/）"
 Cohesion: 0.29
@@ -247,34 +241,50 @@ Nodes (5): assets, document_size, layers, pages, source
 Cohesion: 0.60
 Nodes (4): main(), 递归缩放 result 内全部坐标:{left,top,width,height} 矩形与 {x,y} 点,与原数据同为整数。, scale_coords(), to_data_url()
 
-### Community 36 - "so"
-Cohesion: 0.50
-Nodes (4): Do(), io(), so(), To()
+### Community 36 - "ao"
+Cohesion: 0.33
+Nodes (6): ao(), Do(), io(), oo(), so(), To()
 
 ### Community 37 - "UI 外发版 10.09 图片资源"
 Cohesion: 0.40
 Nodes (4): UI 外发版 10.09 图片资源, 前端素材对应, 文件位置, 重新导出
 
+### Community 38 - "FaceLandmark68Net"
+Cohesion: 0.17
+Nodes (3): FaceLandmark68Net(), PredictSingleAgeAndGenderWithFaceAlignmentTask(), PredictSingleFaceExpressionsWithFaceAlignmentTask()
+
 ### Community 39 - "UI交互界面调整"
 Cohesion: 0.40
 Nodes (4): 1meAI测肤旷视UI问题反馈, UI交互界面调整, 交互, **设计**
 
-### Community 40 - "extractAllFacesAndComputeResults"
-Cohesion: 0.13
-Nodes (9): extendWithAge(), extendWithFaceExpressions(), extendWithGender(), extractAllFacesAndComputeResults(), extractSingleFaceAndComputeResult(), PredictAllAgeAndGenderTask(), PredictAllFaceExpressionsTask(), PredictSingleAgeAndGenderTask() (+1 more)
+### Community 40 - ".predictAgeAndGender"
+Cohesion: 0.28
+Nodes (4): extendWithAge(), extendWithGender(), PredictAllAgeAndGenderTask(), PredictSingleAgeAndGenderTask()
 
-### Community 41 - "initialize"
-Cohesion: 0.29
-Nodes (7): createBrowserEnv(), createFileSystem(), createNodejsEnv(), initialize(), isBrowser(), monkeyPatch(), setEnv()
+### Community 41 - "test/accounts.js"
+Cohesion: 0.22
+Nodes (9): assert, create, fs, http, os, path, ReportUI, SkinScore (+1 more)
 
 ### Community 42 - "export_psd_assets.py"
 Cohesion: 0.70
 Nodes (4): bounds(), layer_at(), main(), save_image()
 
+### Community 43 - ".loadFromDisk"
+Cohesion: 0.29
+Nodes (3): ComposableTask(), getModelUris(), TinyFaceDetector()
+
 ### Community 44 - "start-local.sh"
 Cohesion: 0.50
 Nodes (3): HOST, PORT, start-local.sh script
 
+### Community 47 - "recommendation.js"
+Cohesion: 0.38
+Nodes (6): assert, check(), missing, raw, SS, UI
+
+### Community 48 - "report-persistence.mjs"
+Cohesion: 0.29
+Nodes (6): child, dataRoot, fx, h5, ReportUI, require
+
 ### Community 53 - "护理建议（全量文本，按需求摘抄）"
 Cohesion: 0.10
 Nodes (19): 5\.1 轮廓（紧塑力）, 5\.2 皱纹（紧塑力）, 5\.3 肤色（透亮感）, 5\.4 平滑度（透亮感）, 5\.5 水分（透亮感）, 5\.6 泛红（稳韧力）, 5\.7 痘痘（稳韧力）, 6\.1 轮廓 (+11 more)
@@ -283,21 +293,25 @@ Nodes (19): 5\.1 轮廓（紧塑力）, 5\.2 皱纹（紧塑力）, 5\.3 肤色
 Cohesion: 0.11
 Nodes (17): 1. 结论和依据, 2. 统一比较口径, 3. 逐页差异和调整内容, 4. 报告数据如何接到新 UI, 5. 登录与“我的报告”的真实能力缺口, 6. 原稿中需要整理的内容, 7. 实施顺序与交付物, 8. 验收定义 (+9 more)
 
-### Community 56 - "convLayer"
-Cohesion: 0.15
-Nodes (17): boxPredictionLayer(), convLayer(), convWithBatchNorm(), depthwiseConvLayer(), depthwiseSeparableConv$1(), fullyConnectedLayer(), getStridesForLayerIdx(), leaky() (+9 more)
+### Community 55 - "extractFaceTensors"
+Cohesion: 0.29
+Nodes (3): DetectAllFaceLandmarksTask(), extractFaceTensors(), isTensor3D()
+
+### Community 56 - "report-ui.js"
+Cohesion: 0.90
+Nodes (4): build(), count(), fill(), numeric()
 
 ### Community 57 - "regression.js"
 Cohesion: 0.60
 Nodes (5): diffCtx(), loadJSON(), main(), okNum(), okStr()
 
 ### Community 58 - ".dispose"
-Cohesion: 0.21
-Nodes (11): getSizesForScale(), iou(), NetInput(), nonMaxSuppression(), nonMaxSuppression$1(), normalize$1(), rescaleAndNormalize(), scale() (+3 more)
+Cohesion: 0.16
+Nodes (12): extractBoundingBoxes(), getSizesForScale(), iou(), nonMaxSuppression(), nonMaxSuppression$1(), normalize$1(), rescaleAndNormalize(), scale() (+4 more)
 
-### Community 61 - ".extractWeights"
-Cohesion: 0.09
-Nodes (25): createFaceDetectionNet(), createFaceRecognitionNet(), createMtcnn(), createSsdMobilenetv1(), createTinyFaceDetector(), createTinyYolov2(), extractConvParamsFactory(), extractFCParamsFactory() (+17 more)
+### Community 61 - "NeuralNetwork"
+Cohesion: 0.06
+Nodes (28): createFaceDetectionNet(), createFaceRecognitionNet(), createMtcnn(), createSsdMobilenetv1(), createTinyFaceDetector(), createTinyYolov2(), extractConvParamsFactory(), extractFCParamsFactory() (+20 more)
 
 ### Community 62 - "提交前 Review（2026-10-11）"
 Cohesion: 0.40
@@ -311,52 +325,52 @@ Nodes (4): EXPECTED, fails, pageErrors, pwRoot
 Cohesion: 0.90
 Nodes (4): browser(), init(), read(), request()
 
-### Community 65 - "tt"
-Cohesion: 0.50
-Nodes (4): kn(), tt(), wo(), z()
+### Community 65 - "k"
+Cohesion: 0.13
+Nodes (16): ba(), ca(), fa(), gn(), k(), kn(), pa(), tt() (+8 more)
 
-### Community 66 - "cn"
-Cohesion: 0.67
-Nodes (3): cn(), en(), yn()
+### Community 66 - "wechat-browser.mjs"
+Cohesion: 0.40
+Nodes (4): createAccounts, h5, require, root
 
 ### Community 68 - "analyze-timeout.js"
 Cohesion: 0.20
 Nodes (7): assert, end, fs, path, source, start, vm
 
-### Community 70 - "decodeBoxesLayer"
-Cohesion: 0.67
-Nodes (3): decodeBoxesLayer(), getCenterCoordinatesAndSizesLayer(), outputLayer()
+### Community 71 - "production-loop.mjs"
+Cohesion: 0.50
+Nodes (3): artifacts, errors, samples
 
-### Community 71 - "dr"
-Cohesion: 0.67
-Nodes (3): dr(), na(), ta()
+### Community 77 - "drawFaceExpressions"
+Cohesion: 0.50
+Nodes (3): drawFaceExpressions(), FaceExpressions(), isWithFaceExpressions()
+
+### Community 78 - ".toString"
+Cohesion: 0.50
+Nodes (3): FaceMatch(), la(), pr()
 
 ### Community 79 - "isValidNumber"
-Cohesion: 0.50
-Nodes (4): Dimensions(), isValidNumber(), isValidProbablitiy(), isWithGender()
+Cohesion: 0.33
+Nodes (5): bgrToRgbTensor(), Dimensions(), isValidNumber(), isValidProbablitiy(), isWithGender()
 
 ### Community 80 - "线上微信登录部署记录"
 Cohesion: 0.29
 Nodes (6): 初次部署时的凭据查找记录（现已由用户提供）, 初次部署的线上检查（历史结果）, 后续更新：已配置真实凭据, 已执行, 用户确认加入白名单后的复测, 线上微信登录部署记录
 
-### Community 83 - "Yo"
-Cohesion: 0.67
-Nodes (3): qo(), Xo(), Yo()
-
 ## Knowledge Gaps
-- **173 isolated node(s):** `source`, `document_size`, `pages`, `assets`, `layers` (+168 more)
+- **183 isolated node(s):** `source`, `document_size`, `pages`, `assets`, `layers` (+178 more)
   These have ≤1 connection - possible missing edges or undocumented components.
-- **18 thin communities (<3 nodes) omitted from report** — run `graphify query` to explore isolated nodes.
+- **17 thin communities (<3 nodes) omitted from report** — run `graphify query` to explore isolated nodes.
 
 ## Suggested Questions
 _Questions this graph is uniquely positioned to answer:_
 
-- **Why does `k()` connect `n` to `face-api.js`, `app.js`, `lib/accounts.js`?**
-  _High betweenness centrality (0.083) - this node is a cross-community bridge._
-- **Why does `rate()` connect `lib/accounts.js` to `n`?**
+- **Why does `k()` connect `k` to `face-api.js`, `app.js`, `i`, `n`, `lib/accounts.js`?**
   _High betweenness centrality (0.061) - this node is a cross-community bridge._
-- **Why does `renderReport()` connect `app.js` to `n`?**
-  _High betweenness centrality (0.056) - this node is a cross-community bridge._
+- **Why does `r()` connect `n` to `face-api.js`, `i`, `AgeGenderNet`, `server.js`, `.toString`, `wi`, `C`?**
+  _High betweenness centrality (0.050) - this node is a cross-community bridge._
+- **Why does `renderReport()` connect `app.js` to `k`, `n`?**
+  _High betweenness centrality (0.047) - this node is a cross-community bridge._
 - **Are the 24 inferred relationships involving `n()` (e.g. with `newRecordId()` and `ae()`) actually correct?**
   _`n()` has 24 INFERRED edges - model-reasoned connections that need verification._
 - **Are the 27 inferred relationships involving `r()` (e.g. with `handleExport()` and `handleRecordsQuery()`) actually correct?**
@@ -364,4 +378,4 @@ _Questions this graph is uniquely positioned to answer:_
 - **Are the 7 inferred relationships involving `t()` (e.g. with `renderReport()` and `e()`) actually correct?**
   _`t()` has 7 INFERRED edges - model-reasoned connections that need verification._
 - **What connects `source`, `document_size`, `pages` to the rest of the system?**
-  _173 weakly-connected nodes found - possible documentation gaps or missing edges._
+  _183 weakly-connected nodes found - possible documentation gaps or missing edges._

graphify-out/manifest.json

@@ -40,8 +40,8 @@
     "semantic_hash": ""
   },
   "h5/js/app.js": {
-    "mtime": 1791730674.7447748,
-    "ast_hash": "7fe1cc5b34a9367c3fed5d502d304e2b",
+    "mtime": 1791735440.264282,
+    "ast_hash": "bd18e424efb5c8d037b64d67f869b198",
     "semantic_hash": ""
   },
   "h5/js/head-pose.js": {
@@ -50,13 +50,13 @@
     "semantic_hash": ""
   },
   "h5/js/skin-score.js": {
-    "mtime": 1791645379.1321833,
-    "ast_hash": "12fe3e0e5fd4cbe7f0a79a7da01c380b",
+    "mtime": 1791735501.357809,
+    "ast_hash": "783b6c9abbe2bf1832e8eafe5aead7e3",
     "semantic_hash": ""
   },
   "h5/server.js": {
-    "mtime": 1791710640.6026413,
-    "ast_hash": "a63e26c0f78becff09cc50e427cb0054",
+    "mtime": 1791735707.6717124,
+    "ast_hash": "6bdce03487aaaf9e3ab62f6b39f7a418",
     "semantic_hash": ""
   },
   "h5/test/build_fixtures.py": {
@@ -80,8 +80,8 @@
     "semantic_hash": ""
   },
   "h5/test/regression.js": {
-    "mtime": 1790528158.6705542,
-    "ast_hash": "d83fc0a378d1117fc2203569d325ca24",
+    "mtime": 1791735501.3654072,
+    "ast_hash": "4a712e295c115b6051726cd028e67be2",
     "semantic_hash": ""
   },
   "h5/vendor/faceapi/face-api.js": {
@@ -140,8 +140,8 @@
     "semantic_hash": ""
   },
   "h5/index.html": {
-    "mtime": 1791646269.975195,
-    "ast_hash": "c2661595e503e2f166f1be390696ce0c",
+    "mtime": 1791735440.2604818,
+    "ast_hash": "444fca45ad900710607cb15b852c0305",
     "semantic_hash": ""
   },
   "h5/vendor/faceapi/README.md": {
@@ -815,8 +815,8 @@
     "semantic_hash": ""
   },
   "h5/js/admin.js": {
-    "mtime": 1791556943.863973,
-    "ast_hash": "573a1a971cc3d00a16c8b9d53529f243",
+    "mtime": 1791735707.6717124,
+    "ast_hash": "0de3075fb017762401fd2d27eb26abcb",
     "semantic_hash": ""
   },
   "h5/js/region-data.js": {
@@ -860,8 +860,8 @@
     "semantic_hash": ""
   },
   "h5/admin.html": {
-    "mtime": 1791556943.863973,
-    "ast_hash": "8dad311cd83b0ec640bcf71ddb9a255c",
+    "mtime": 1791735369.9312572,
+    "ast_hash": "ddfe588d49452e3973e21b080e61dbd7",
     "semantic_hash": ""
   },
   "h5/vendor/mobile-select/README.md": {
@@ -1245,23 +1245,23 @@
     "semantic_hash": ""
   },
   "h5/js/report-ui.js": {
-    "mtime": 1791646198.024173,
-    "ast_hash": "5cb7f2ac736b6859ed580f5c934b1061",
+    "mtime": 1791735369.923658,
+    "ast_hash": "b05b0a2181711ce6f05bbe6c16179170",
     "semantic_hash": ""
   },
   "h5/lib/accounts.js": {
-    "mtime": 1791710640.6026413,
-    "ast_hash": "e36024ccb6c9d6a2d2a0187942d5c65b",
+    "mtime": 1791735369.9274576,
+    "ast_hash": "1a0677e439938aa2e9e4decf38fad568",
     "semantic_hash": ""
   },
   "h5/test/accounts.js": {
-    "mtime": 1791710640.6026413,
-    "ast_hash": "f036faf880172eb33e3d0fd4e5c462f7",
+    "mtime": 1791735546.0452847,
+    "ast_hash": "aa05cb7d924b1ddbe8834faeab24588f",
     "semantic_hash": ""
   },
   "h5/test/report-persistence.mjs": {
-    "mtime": 1791710640.6026413,
-    "ast_hash": "92699686424262cfe208355ce040f236",
+    "mtime": 1791735750.0753446,
+    "ast_hash": "2933add99f8983a590608ba203182096",
     "semantic_hash": ""
   },
   "h5/test/ui-20261009.mjs": {
@@ -1280,8 +1280,8 @@
     "semantic_hash": ""
   },
   "docs/10.09\u65b0\u7248H5\u5b9e\u65bd\u4e0e\u9a8c\u6536.md": {
-    "mtime": 1791710850.0228443,
-    "ast_hash": "e3126d7bce54a235d580778609a46dd3",
+    "mtime": 1791735707.6755157,
+    "ast_hash": "6e83c0cd6df4b711c9c526a83f1c33a5",
     "semantic_hash": ""
   },
   "docs/UI\u5bf9\u9f5010.09\u8bbe\u8ba1\u7a3f-\u8c03\u7814\u4e0e\u5b9e\u65bd\u8ba1\u5212.md": {
@@ -1563,5 +1563,15 @@
     "mtime": 1791730745.132436,
     "ast_hash": "04095d09dce4e81349b9256a0e8cb540",
     "semantic_hash": ""
+  },
+  "h5/test/recommendation.js": {
+    "mtime": 1791735559.0619977,
+    "ast_hash": "589f7046d94485dfc1df20a6129640e0",
+    "semantic_hash": ""
+  },
+  "docs/\u4ea7\u54c1\u63a8\u8350\u89c4\u5219-\u80a4\u9f84\u7248.md": {
+    "mtime": 1791735707.6717124,
+    "ast_hash": "69c1278c6c8eea5866982940e20e421c",
+    "semantic_hash": ""
   }
 }

h5/admin.html

@@ -44,7 +44,7 @@
           <option>NETWORK_ERROR</option>
         </select>
       </label>
-      <label>年龄 <input type="number" id="f-agemin" min="1" max="120" placeholder="最小"> ~ <input type="number" id="f-agemax" min="1" max="120" placeholder="最大"></label>
+      <label>实际年龄 <input type="number" id="f-agemin" min="1" max="120" placeholder="最小"> ~ <input type="number" id="f-agemax" min="1" max="120" placeholder="最大"></label>
       <label>产品组别
         <select id="f-group">
           <option value="">全部</option>
@@ -55,12 +55,12 @@
         <select id="f-dim">
           <option value="">不限</option>
           <option value="firm">紧塑力</option>
-          <option value="vita">活力感</option>
-          <option value="stab">稳定力</option>
+          <option value="vita">透亮感</option>
+          <option value="stab">稳韧力</option>
         </select>
         <select id="f-dimlevel">
           <option value="">不限档位</option>
-          <option>预警</option><option>改善</option><option>良好</option>
+          <option value="预警">改善</option><option value="改善">良好</option><option value="良好">优秀</option>
         </select>
       </label>
       <label>记录ID <input type="text" id="f-q" placeholder="精确/前缀匹配"></label>
@@ -72,7 +72,7 @@
         <thead>
           <tr>
             <th>照片</th><th>记录ID</th><th>年龄/性别/地区/肤质</th><th>拍摄状态</th><th>上传状态</th>
-            <th>检测状态</th><th>总分</th><th>三维度档位</th><th>推荐精华</th><th>耗时</th><th>创建时间</th>
+            <th>检测状态</th><th>总分</th><th>肌龄（原始 / 展示）</th><th>三维度评分 / 档位</th><th>推荐精华</th><th>耗时</th><th>创建时间</th>
           </tr>
         </thead>
         <tbody></tbody>
@@ -149,6 +149,8 @@
 
 <div id="toast" class="toast hidden"></div>
 
+<script src="js/report-copy.js"></script>
+<script src="js/report-ui.js"></script>
 <script src="js/admin.js"></script>
 </body>
 </html>

h5/css/admin.css

@@ -62,7 +62,7 @@ tbody tr:last-child td { border-bottom: none; }
 .mono { font-family: ui-monospace, Consolas, monospace; font-size: 12px; color: #4a5568; }
 .sub { color: #8a94a6; font-size: 12px; }
 
-/* 档位标签:预警=红 / 改善=橙黄 / 良好=绿 */
+/* 原始档位样式键不变；展示按 ReportUI 映射为改善 / 良好 / 优秀。 */
 .lv {
   display: inline-block; padding: 2px 8px; border-radius: 10px; font-size: 12px; margin: 1px 2px;
 }

h5/index.html

@@ -261,6 +261,7 @@
 
     <!-- 产品推荐大标题独立于白卡之外,直接落在灰底上(Figma 08) -->
     <h2 class="prod-title">感受高光时刻<br>从日常的好状态开始</h2>
+    <p id="product-unavailable" class="care-item" hidden>本次未获取有效肌龄，暂无法推荐产品，建议重新检测。</p>
     <div class="card card-product">
       <div class="prod-card">
         <div class="prod-info">

h5/js/admin.js

@@ -24,7 +24,7 @@ function fmtMs(ms) { return ms == null ? '-' : ms + 'ms'; }
 function fmtScore(v) { return v == null ? '-' : v; }
 
 var LV_CLASS = { '预警': 'lv-bad', '改善': 'lv-warn', '良好': 'lv-good' };
-var DIM_NAMES = { firm: '紧塑力', vita: '活力感', stab: '稳定力' };
+var DIM_NAMES = ReportUI.dimNames;
 var CAT_ORDER = ['contour', 'wrinkles', 'smooth', 'tone', 'sens', 'acne'];
 var ST_CLASS = { '成功': 'st-ok', '失败': 'st-fail', '检测中': 'st-ing', '待检测': 'st-wait' };
 
@@ -60,15 +60,16 @@ function switchTab(name) {
 /* ─── 档位标签 ─── */
 function lvHtml(level) {
   if (!level) return '<span class="sub">-</span>';
-  return '<span class="lv ' + (LV_CLASS[level] || '') + '">' + esc(level) + '</span>';
+  return '<span class="lv ' + (LV_CLASS[level] || '') + '">' + esc(ReportUI.levelMap[level] || level) + '</span>';
 }
 function dimsHtml(dims) {
   if (!dims) return '<span class="sub">-</span>';
   var h = '';
   for (var k in DIM_NAMES) {
     var d = dims[k];
-    h += '<span class="lv ' + (d && LV_CLASS[d.level] || '') + '" title="' + DIM_NAMES[k] + '">' +
-      DIM_NAMES[k].slice(0, 1) + (d ? esc(d.level) : '-') + '</span>';
+    h += '<div><span class="lv ' + (d && LV_CLASS[d.level] || '') + '" title="' + DIM_NAMES[k] + '">' +
+      DIM_NAMES[k] + ' ' + (d && typeof d.S === 'number' ? d.S.toFixed(1) : '-') + ' ' +
+      (d ? esc(ReportUI.levelMap[d.level] || '-') : '-') + '</span></div>';
   }
   return h;
 }
@@ -80,6 +81,19 @@ function profileHtml(it) {
   parts.push(it.daily_skin_type || '-');
   return esc(parts.join(' / '));
 }
+function ageText(age) { return age == null ? '-' : esc(age) + '岁'; }
+function skinAgeHtml(r) {
+  return '<b>原始 ' + ageText(r.skin_age) + '</b><div class="sub">展示 ' + ageText(r.skin_age_display) + '</div>';
+}
+function productName(name) {
+  for (var i = 0; i < ReportUI.products.length; i++) {
+    var p = ReportUI.products[i];
+    if (p.match === name) return p.title + '（' + ['25+', '35+', '50+'][i] + '）';
+  }
+  return name || '-';
+}
+function ruleLabel(r) { return r.recommendation ? r.recommendation.version : '旧版规则（历史记录）'; }
+function tierText(tier) { return tier ? ['25+ / I', '35+ / II', '50+ / III'][tier - 1] : '无'; }
 
 /* ═══════════ 记录列表 ═══════════ */
 var pageState = { page: 1, page_size: 20, total: 0 };
@@ -108,7 +122,7 @@ function loadRecords() {
     var tbody = $('#tbl-records tbody');
     tbody.innerHTML = '';
     if (!data.items.length) {
-      tbody.innerHTML = '<tr><td colspan="11" style="text-align:center;color:#8a94a6;padding:32px">暂无记录</td></tr>';
+      tbody.innerHTML = '<tr><td colspan="12" style="text-align:center;color:#8a94a6;padding:32px">暂无记录</td></tr>';
     }
     for (var i = 0; i < data.items.length; i++) {
       (function (it) {
@@ -126,8 +140,9 @@ function loadRecords() {
           '<td><span class="st ' + (ST_CLASS[it.detect_status] || '') + '">' + esc(it.detect_status) + '</span>' +
             (it.fail_code ? '<div class="sub">' + esc(it.fail_code) + '</div>' : '') + '</td>' +
           '<td><b>' + fmtScore(it.total_score) + '</b></td>' +
+          '<td>' + skinAgeHtml(it) + '</td>' +
           '<td>' + dimsHtml(it.dims) + '</td>' +
-          '<td>' + esc(it.essence || '-') + '</td>' +
+          '<td>' + esc(productName(it.essence)) + '<div class="sub">' + esc(ruleLabel(it)) + '</div></td>' +
           '<td class="mono">' + fmtMs(it.cost_ms) + '</td>' +
           '<td class="mono">' + fmtTime(it.created_at) + '</td>';
         tr.addEventListener('click', function (e) {
@@ -230,7 +245,7 @@ function loadStats() {
       for (var l = 0; l < levels.length; l++) {
         var cnt = dist[levels[l][0]] || 0;
         bars += '<div style="display:flex;align-items:center;gap:8px;margin:3px 0">' +
-          '<span style="width:34px;color:#5a6578;font-size:12px">' + levels[l][0] + '</span>' +
+          '<span style="width:34px;color:#5a6578;font-size:12px">' + ReportUI.levelMap[levels[l][0]] + '</span>' +
           '<span style="flex:1;height:14px;background:#f0f2f5;border-radius:3px;overflow:hidden">' +
           '<i style="display:block;height:100%;width:' + (cnt * 100 / max) + '%" class="' + levels[l][1] + '"></i></span>' +
           '<span style="width:24px;text-align:right;font-size:12px;color:#4a5568">' + cnt + '</span></div>';
@@ -342,7 +357,7 @@ function renderDetail(data) {
 
   /* 档案 */
   h += dSec('档案信息', '<div class="d-kv">' +
-    '<span>年龄 <b>' + (r.real_age != null ? r.real_age + '岁' : '-') + '</b></span>' +
+    '<span>实际年龄 <b>' + ageText(r.real_age) + '</b></span>' +
     '<span>性别 <b>' + esc(r.gender || '-') + '</b></span>' +
     '<span>地区 <b>' + esc(r.region || '-') + '</b></span>' +
     '<span>日常肤质 <b>' + esc(r.daily_skin_type || '-') + '</b></span>' +
@@ -374,20 +389,22 @@ function renderDetail(data) {
   if (r.detect_status === '成功') {
     h += dSec('综合', '<div class="d-kv">' +
       '<span>总分 <b>' + fmtScore(r.total_score) + '</b></span>' +
-      '<span>肌肤年龄(接口原始) <b>' + fmtScore(r.skin_age) + '岁</b></span>' +
+      '<span>原始肌龄（新推荐依据） <b>' + ageText(r.skin_age) + '</b></span>' +
+      '<span>前端展示肌龄 <b>' + ageText(r.skin_age_display) + '</b></span>' +
       '<span>六分类均分S <b>' + fmtScore(r.avg_s) + '</b></span>' +
       '<span>左下颌线角 <b>' + (r.jawline_angle_l != null ? r.jawline_angle_l + '°' : '-') + '</b></span>' +
       '<span>右下颌线角 <b>' + (r.jawline_angle_r != null ? r.jawline_angle_r + '°' : '-') + '</b></span>' +
       '</div>');
+    h += '<p class="sub">前端展示肌龄沿用实际年龄±3岁的修正；新推荐只使用原始肌龄。历史报告保留当时的推荐。</p>';
 
     /* 6 分类表格 */
     var catRows = '';
     for (var i = 0; i < CAT_ORDER.length; i++) {
       var ck = CAT_ORDER[i], cat = r.cats && r.cats[ck];
-      var catName = (cat && cat.name) || ({ contour: '轮廓', wrinkles: '皱纹', smooth: '光滑', tone: '肤色水润', sens: '敏感', acne: '痘痘' })[ck];
+      var catName = ({ contour: '轮廓', wrinkles: '皱纹', smooth: '平滑度', tone: '肤色 / 水分', sens: '泛红', acne: '痘痘' })[ck];
       catRows += '<tr><td>' + esc(catName) + '</td><td>' + fmtScore(cat && cat.P) + '</td><td>' + fmtScore(cat && cat.S) + '</td><td>' + lvHtml(cat && cat.level) + '</td></tr>';
     }
-    h += dSec('六分类评分', '<table class="d-table"><thead><tr><th>分类</th><th>P</th><th>S</th><th>档位</th></tr></thead><tbody>' + catRows + '</tbody></table>');
+    h += dSec('原始六分类评分（肤色 / 水分共用一项）', '<table class="d-table"><thead><tr><th>分类</th><th>P</th><th>S</th><th>档位</th></tr></thead><tbody>' + catRows + '</tbody></table>');
 
     /* 3 维度表格 */
     var dimRows = '';
@@ -399,7 +416,13 @@ function renderDetail(data) {
 
     /* 文案全文 */
     var copyHtml = '';
-    if (r.copy_6) {
+    var presentation = r.presentation;
+    if (presentation && presentation.views) {
+      for (var v = 0; v < presentation.views.length; v++) {
+        var view = presentation.views[v];
+        copyHtml += '<div class="d-copy"><b>' + esc(view.name) + ' / ' + esc(view.level || '未获取') + '：</b>' + esc(view.text) + '</div>';
+      }
+    } else if (r.copy_6) {
       for (var c2 = 0; c2 < CAT_ORDER.length; c2++) {
         var txt = r.copy_6[CAT_ORDER[c2]];
         if (txt) copyHtml += '<div class="d-copy"><b>' + esc((r.cats[CAT_ORDER[c2]] || {}).name || '') + ':</b>' + esc(txt) + '</div>';
@@ -407,19 +430,33 @@ function renderDetail(data) {
     }
     if (r.copy_summary) copyHtml += '<div class="d-copy"><b>综合文案:</b>' + esc(r.copy_summary) + '</div>';
     if (r.lifestyle) copyHtml += '<div class="d-copy"><b>生活方式:</b>' + esc(r.lifestyle) + '</div>';
-    if (copyHtml) h += dSec('生成文案', copyHtml);
+    if (copyHtml) h += dSec(presentation ? '前端报告结论（保存快照）与计算摘要' : '旧版生成文案（无前端快照）', copyHtml);
 
-    if (r.careAdvice && r.careAdvice.length) {
+    var care = presentation ? presentation.care : r.careAdvice;
+    if (care && care.length) {
       var advice = '';
-      for (var a = 0; a < r.careAdvice.length; a++) {
-        advice += '<div class="d-copy"><b>[' + esc(r.careAdvice[a].tag) + ']</b>' + esc(r.careAdvice[a].text) + '</div>';
+      for (var a = 0; a < care.length; a++) {
+        advice += '<div class="d-copy"><b>[' + esc(care[a].name || care[a].tag) + ']</b>' + esc(care[a].text) + '</div>';
       }
       h += dSec('护理建议', advice);
     }
 
-    h += dSec('推荐精华', '<div class="d-copy"><b>' + esc((r.essence && r.essence.name) || '-') + '</b><br>' +
-      esc((r.essence && r.essence.text) || '') + '</div>' +
-      '<div class="d-kv"><span>触发判据 <b>' + esc(r.trigger || '-') + '</b></span></div>');
+    var product = presentation && presentation.product;
+    h += dSec('推荐精华', '<div class="d-copy"><b>' + esc(productName(r.essence && r.essence.name)) + '</b><br>' +
+      esc(product ? product.name + '双仓精华液' : (r.essence && r.essence.text) || '') + '</div>' +
+      '<div class="d-kv"><span>规则版本 <b>' + esc(ruleLabel(r)) + '</b></span>' +
+      '<span>触发判据 <b>' + esc(r.trigger || '-') + '</b></span></div>');
+    if (r.recommendation) {
+      var rec = r.recommendation, scoreParts = [];
+      for (var sk in DIM_NAMES) {
+        var score = rec.scores[sk];
+        scoreParts.push(DIM_NAMES[sk] + '：' + (score == null ? '未获取' : String(score)));
+      }
+      h += dSec('推荐依据', '<div class="d-kv"><span>使用原始肌龄 <b>' + ageText(rec.skinAge) + '</b></span>' +
+        '<span>基础档 <b>' + esc(tierText(rec.baseTier)) + '</b></span>' +
+        '<span>最终档 <b>' + esc(tierText(rec.tier)) + '</b></span></div>' +
+        '<div class="d-copy">' + esc(rec.reason) + '</div><div class="sub">判定分（未舍入S，不使用前端一位小数）：' + esc(scoreParts.join('；')) + '</div>');
+    }
 
     if (r.notes && r.notes.length) {
       var notes = '<ul class="d-notes">';

h5/js/app.js

@@ -1194,6 +1194,7 @@
     var product = presentation.product;
     $('.card-product').hidden = !product;
     $('.prod-title').hidden = !product;
+    $('#product-unavailable').hidden = !!product;
     if (product) {
       $('#product-title').textContent = product.title;
       $('#product-name').textContent = product.name + '双仓精华液';

h5/js/report-ui.js

@@ -15,6 +15,7 @@
     { tier: 3, match: '50+重塑精华', title: "1'me 主场精华 III", name: '以我从容有态构塑', image: 'assets/img/ui-20261009/product-50.png' }
   ];
   var levelMap = { '预警': '改善', '改善': '良好', '良好': '优秀' };
+  var dimNames = { firm: '紧塑力', vita: '透亮感', stab: '稳韧力' };
   function numeric(value) { return typeof value === 'number' && isFinite(value); }
   function fill(template, vars) {
     return template.replace(/\{([a-z_]+)\}/g, function(_, key) {
@@ -66,7 +67,7 @@
     var chosen = views.filter(function(v) { return ranked.indexOf(v.source) >= 0 && v.advice; });
     var care = chosen.map(function(v) { return { name: v.name, text: v.advice.care + v.advice.lifestyle }; });
     var product = products.filter(function(p) { return p.match === vm.essence.name; })[0] || null;
-    return { version: Copy.version, dims: { firm: '紧塑力', vita: '透亮感', stab: '稳韧力' }, views: views, care: care, product: product };
+    return { version: Copy.version, dims: dimNames, views: views, care: care, product: product };
   }
-  return { build: build, products: products, levelMap: levelMap };
+  return { build: build, products: products, levelMap: levelMap, dimNames: dimNames };
 });

h5/js/skin-score.js

@@ -1,5 +1,6 @@
 /* ═════ 巨子测肤 H5 — 三维度×6分类评分引擎（纯逻辑，浏览器/Node 双端） ═════
- * 最高契约：docs/三维度评分与话术生成需求文档_V2.2.docx（公式/锁值/话术/产品推荐以其为准）
+ * 评分契约：docs/三维度评分与话术生成需求文档_V2.2.docx（公式/锁值/原始话术）
+ * 推荐覆盖契约：docs/产品推荐规则-肤龄版.md（2026-10-12 用户确认）
  * 浏览器：window.SkinScore；Node：module.exports。ES5、零依赖、不碰 DOM。
  *
  * 字段口径（已对 reference/facepp-skinanalyze/examples/ 真实 pro JSON 复核）：
@@ -63,7 +64,7 @@
   function round2(x) { return Math.round((x + 1e-9) * 100) / 100; } // 1e-9 修正 9.445 类浮点
   function sRaw(P) { return 7 + 3 * P / 100; }
   function sShow(P) { // 展示层钳制：P 可为负（肤色扣分/大面积敏感），S 会 <7，与文档「S 7.0–10.0」口径矛盾；
-    if (P == null) return null;   // 仅钳 viewModel 展示值，判档/最差项/精华三重判据仍用未舍入原值
+    if (P == null) return null;   // 仅钳展示值，推荐边界比较仍用未舍入维度 S 原值
     var s = sRaw(P);
     return round2(s < 7 ? 7 : (s > 10 ? 10 : s));
   }
@@ -589,6 +590,47 @@
     3: { name: '50+重塑精华', text: '建议搭配使用50+重塑精华：针对熟龄肌松弛与轮廓流失问题，深层重塑支撑结构，改善明显松弛与纹路，重建面部紧实与弹性。' }
   };
 
+  // 2026-10-12 已确认的业务规则。只接收原始肤龄和三维度未舍入 S，
+  // 不使用实际年龄、展示肤龄、六分类均分或最差分类；历史快照不重算。
+  function recommendProduct(skinAge, scores) {
+    scores = scores || {};
+    var firm = isNum(scores.firm) ? scores.firm : null;
+    var vita = isNum(scores.vita) ? scores.vita : null;
+    var stab = isNum(scores.stab) ? scores.stab : null;
+    var rec = { version: '2026-10-12-skin-age-v1', skinAge: skinAge,
+      baseTier: null, tier: null, boundary: null, adjustment: 'none',
+      scores: { firm: firm, vita: vita, stab: stab }, reason: '', triggers: [] };
+    if (!isNum(skinAge) || skinAge <= 0) {
+      rec.adjustment = 'unavailable';
+      rec.reason = '未获取有效原始肌龄，暂不推荐具体产品；不以实际年龄代替。';
+      return rec;
+    }
+    rec.baseTier = skinAge < 35 ? 1 : (skinAge < 50 ? 2 : 3);
+    rec.tier = rec.baseTier;
+    rec.triggers.push('肤龄基础分组');
+    if (skinAge >= 32 && skinAge <= 38) rec.boundary = 35;
+    else if (skinAge >= 47 && skinAge <= 53) rec.boundary = 50;
+    if (rec.boundary != null && skinAge < rec.boundary && firm != null && firm < 8) {
+      rec.tier++;
+      rec.adjustment = 'up';
+      rec.reason = '处于' + rec.boundary + '岁下侧边界，紧塑力低于8分，升一档。';
+      rec.triggers.push(rec.boundary + '岁边界低分升档');
+    } else if (rec.boundary != null && skinAge >= rec.boundary && firm != null && firm >= 9
+        && ((vita != null && vita >= 9) || (stab != null && stab >= 9))) {
+      rec.tier--;
+      rec.adjustment = 'down';
+      rec.reason = '处于' + rec.boundary + '岁上侧边界，紧塑力≥9且透亮感或稳韧力≥9，降一档。';
+      rec.triggers.push(rec.boundary + '岁边界高分降档');
+    } else if (skinAge < 32) {
+      rec.reason = '原始肌龄小于32岁，固定推荐25+，轮廓或皱纹低分不升档。';
+    } else if (rec.boundary != null) {
+      rec.reason = '处于' + rec.boundary + '岁边界，未满足升降档条件，保持肤龄基础档。';
+    } else {
+      rec.reason = '不在35±3或50±3岁边界内，仅按原始肌龄推荐。';
+    }
+    return rec;
+  }
+
   /* ─── V2.3 §8 护理建议 ─── */
   var SKIN_TYPE_TEXT = { // §8.6 五类映射【肤质定制】段全文 + 联动句
     '干性肌肤': { base: '你的日常肤质为干性肌肤：全脸皮脂分泌偏少，角质层保水能力弱，易紧绷起屑。护理以高效保湿与封闭锁水为核心，洁面选用温和低泡产品，避免频繁去角质，面霜可选择含神经酰胺、角鲨烷等封闭性成分的配方。',
@@ -651,14 +693,12 @@
     var skinAge = val(r.skin_age);
     var skinAgeDisplay = skinAge, ageClampNote = '';
     if (realAge == null) {
-      notes.push('未提供实际年龄：产品组别、肌肤年龄钳制与精华「年龄组档」判据均不生效');
+      notes.push('未提供实际年龄：不修正展示肌龄；产品推荐仍使用接口原始肌龄');
     } else if (skinAge != null) {
       skinAgeDisplay = Math.min(Math.max(skinAge, realAge - 3), realAge + 3);
       if (skinAgeDisplay !== skinAge) ageClampNote = '触发(原' + skinAge + '→' + skinAgeDisplay + ')';
     }
 
-    var group = null;
-    if (realAge != null) group = realAge < 32 ? '25+组' : (realAge <= 45 ? '35+组' : '50+组');
     var skinTypeCn = skinTypeCnOf(r, profile); // {skin_type_cn}：接口枚举优先，回退用户自选肤质
 
     // 6 分类
@@ -707,25 +747,16 @@
       if (!worst || cP < worst.P) worst = { catKey: def.key, name: def.name, P: cP, S: sShow(cP) };
     }
 
-    // 精华推荐：三重判据取最强档，triggers 记录命中（档位=最终档）的判据
-    var crits = []; // {name, tier}
-    if (avgRaw != null) crits.push({ name: '均分档', tier: avgRaw >= 9 ? 1 : (avgRaw >= 8 ? 2 : 3) });
-    if (worst) {
-      var wS = sRaw(worst.P);
-      if (wS < 7) crits.push({ name: '最差项约束', tier: 3 });
-      else if (wS < 8) crits.push({ name: '最差项约束', tier: 2 });
-    }
-    if (group) crits.push({ name: '年龄组档', tier: group === '25+组' ? 1 : (group === '35+组' ? 2 : 3) });
-    var tier = 1;
-    if (crits.length) {
-      tier = crits[0].tier;
-      for (i = 0; i < crits.length; i++) tier = Math.max(tier, crits[i].tier); // 推荐强度只升不降
-    } else {
-      notes.push('三重判据均不可用，精华推荐缺省为 25+束紧精华');
-    }
-    var triggers = [];
-    for (i = 0; i < crits.length; i++) if (crits[i].tier === tier) triggers.push(crits[i].name);
-    var essence = { name: ESSENCE[tier].name, text: ESSENCE[tier].text, triggers: triggers };
+    var recommendation = recommendProduct(skinAge, {
+      firm: dims.firm.P == null ? null : sRaw(dims.firm.P),
+      vita: dims.vita.P == null ? null : sRaw(dims.vita.P),
+      stab: dims.stab.P == null ? null : sRaw(dims.stab.P)
+    });
+    var tier = recommendation.tier, triggers = recommendation.triggers;
+    var group = tier == null ? null : ['25+组', '35+组', '50+组'][tier - 1];
+    var essence = { name: tier == null ? '暂无法推荐' : ESSENCE[tier].name,
+      text: tier == null ? recommendation.reason : ESSENCE[tier].text, tier: tier, triggers: triggers };
+    if (tier == null) notes.push(recommendation.reason);
 
     // 近期优先关注：最差维度
     var focus = { name: '', tip: '' }, fP = null;
@@ -764,7 +795,7 @@
       ], '，');
       summary = '【综合】' + (head ? head + '；' : '')
         + '六分类S均分' + fmtS(avgRaw) + (worst ? '、最差分类' + worst.name + ' S=' + fmtS(sRaw(worst.P)) : '') + '。'
-        + '按三重判据' + (triggers.length ? '（触发：' + triggers.join('、') + '）' : '') + '推荐' + essence.name
+        + (tier == null ? recommendation.reason : '按肤龄及边界规则（触发：' + triggers.join('、') + '）推荐' + essence.name)
         + (improve ? '：' + improve : '。')
         + (tail ? tail + '。' : '');
     }
@@ -795,6 +826,7 @@
       cats: cats,
       group: group,
       essence: essence,
+      recommendation: recommendation,
       avgS: avgS,
       worst: worst,
       focus: focus,
@@ -868,6 +900,7 @@
 
   return {
     buildViewModel: buildViewModel,
+    recommendProduct: recommendProduct,
     MOCK_RESULT: MOCK_RESULT,
     MOCK_SIDE: MOCK_SIDE
   };
@@ -909,14 +942,14 @@ if (typeof require === 'function' && typeof module !== 'undefined' && require.ma
     ok(vm.cats.wrinkles.S === 7.69 && vm.cats.smooth.S === 9.45 && vm.cats.tone.S === 9.45, 'S=7+3×P/100 两位小数（含 9.445→9.45 浮点修正）');
     ok(vm.dims.firm.P === 23 && vm.dims.vita.P === 81.5 && vm.dims.stab.P === 85, '维度木桶 min：紧塑力23 / 活力感81.5 / 稳定力85');
     ok(vm.skinAgeRaw === 20 && vm.skinAgeDisplay === 22 && vm.ageClampNote === '触发(原20→22)', '肌肤年龄钳制 20→22 留痕，原值不改写');
-    ok(vm.group === '25+组', 'realAge=25 → 25+组');
-    ok(vm.essence.name === '35+支撑精华' && vm.essence.triggers.length === 1 && vm.essence.triggers[0] === '最差项约束',
-      '精华三重判据：均分9.32→25+、最差S7.69→至少35+、年龄→25+，取最强档35+，triggers=["最差项约束"]');
+    ok(vm.group === '25+组', '原始肌龄20 → 25+组');
+    ok(vm.essence.name === '25+束紧精华' && vm.recommendation.tier === 1,
+      '肌龄小于32，即使皱纹低分也固定25+，不再使用最差项升档');
     ok(vm.avgS === 9.32, '六分类S均分 9.32');
     ok(vm.worst.catKey === 'wrinkles' && vm.worst.P === 23, '最差分类 = 皱纹');
     ok(vm.focus.name === '紧塑力', '近期优先关注 = 最差维度紧塑力');
     ok(vm.total === 79, '综合评分 = score_info.total_score');
-    ok(vm.summary.indexOf('【综合】') === 0 && vm.summary.indexOf('35+支撑精华') > 0, '综合文案生成');
+    ok(vm.summary.indexOf('【综合】') === 0 && vm.summary.indexOf('25+束紧精华') > 0, '综合文案生成');
     ok(vm.cats.smooth.text.indexOf('粗大毛孔296颗、黑头39颗') >= 0 && vm.cats.smooth.text.indexOf('纹理分87') >= 0,
       '光滑话术含粗大毛孔/黑头/纹理分变量（分区计数 4 区求和 296）');
     checkTexts(vm, 'T0');
@@ -939,7 +972,7 @@ if (typeof require === 'function' && typeof module !== 'undefined' && require.ma
     ok(vm2.dims.firm.P === Math.min(vm2.cats.contour.P, vm2.cats.wrinkles.P)
       && vm2.dims.vita.P === Math.min(vm2.cats.smooth.P, vm2.cats.tone.P)
       && vm2.dims.stab.P === Math.min(vm2.cats.sens.P, vm2.cats.acne.P), '三维度 P = min(旗下两分类)');
-    ok(vm2.group === '35+组', 'realAge=40 → 35+组');
+    ok(vm2.group === '25+组', '实际年龄40不影响推荐，原始肌龄20仍为25+组');
 
     console.log('== T3 钳制边界 ==');
     var r3 = clone(SS.MOCK_RESULT);
@@ -950,7 +983,7 @@ if (typeof require === 'function' && typeof module !== 'undefined' && require.ma
     var vm3b = SS.buildViewModel(r3, SS.MOCK_SIDE, { realAge: 25 });
     ok(vm3b.skinAgeDisplay === 28 && vm3b.ageClampNote === '触发(原40→28)' && vm3b.skinAgeRaw === 40, '上边界钳制 40→28');
 
-    console.log('== T4 精华三重判据：最差项 S<7 → 50+重塑 ==');
+    console.log('== T4 最差项不再覆盖肤龄推荐 ==');
     var r4 = clone(SS.MOCK_RESULT);
     r4.score_info.melanin_score = 0;
     r4.score_info.water_score = 0;
@@ -959,8 +992,8 @@ if (typeof require === 'function' && typeof module !== 'undefined' && require.ma
     r4.water = { water_score: 0, water_area: 40, water_severity: 70 }; // sev≥60 但 score<80，不触发换算
     var vm4 = SS.buildViewModel(r4, SS.MOCK_SIDE, { realAge: 25 });
     ok(approx(vm4.cats.tone.P, -7.5), '肤色水润P = 0.5×(0−10−5)+0.5×0 = −7.5（P 原值不动，不夹逼）');
-    ok(vm4.cats.tone.S === 7, '展示层 S 钳到 7（未钳原值 6.78 仍参与判档）');
-    ok(vm4.essence.name === '50+重塑精华' && vm4.essence.triggers.indexOf('最差项约束') >= 0, '最差项 S 原值<7 → 50+重塑精华，triggers 含最差项约束（三重判据用未舍入原值）');
+    ok(vm4.cats.tone.S === 7, '展示层 S 钳到7，评分公式不变');
+    ok(vm4.essence.name === '25+束紧精华', '原始肌龄20，即使最差项S<7也不升档');
     checkTexts(vm4, 'T4');
 
     console.log('== T5 缺数据降级（V2.2 第十章异常与边界） ==');
@@ -977,9 +1010,9 @@ if (typeof require === 'function' && typeof module !== 'undefined' && require.ma
     var vm5b = SS.buildViewModel(clone(SS.MOCK_RESULT), null, { realAge: 25 });
     ok(vm5b.cats.contour.P === 100 && vm5b.notes.join('|').indexOf('侧脸') >= 0, '无侧脸 → 轮廓 P 由眼袋单独决定并记 note');
     var vm5c = SS.buildViewModel(clone(SS.MOCK_RESULT), SS.MOCK_SIDE, {});
-    ok(vm5c.group === null && vm5c.skinAgeDisplay === 20 && vm5c.ageClampNote === ''
+    ok(vm5c.group === '25+组' && vm5c.skinAgeDisplay === 20 && vm5c.ageClampNote === ''
       && vm5c.essence.triggers.indexOf('年龄组档') < 0 && vm5c.notes.join('|').indexOf('实际年龄') >= 0,
-      'realAge 缺失 → group=null、不钳制、判据③不生效并记 note');
+      '实际年龄缺失不钳制，但原始肌龄仍可推荐25+');
     checkTexts(vm5, 'T5');
 
     console.log('== T6 下颌线 quality 降级 ==');
@@ -1138,8 +1171,8 @@ if (typeof require === 'function' && typeof module !== 'undefined' && require.ma
       ok(vm.cats.sens.P === 99 && vm.cats.acne.P === 70 && vm.dims.stab.P === 70 && vm.dims.stab.S === 9.1,
         '敏感 99 / 痘痘 70（min(分级70, 100−12)），稳定力 70 / S=9.1');
       ok(vm.avgS === 9.05 && vm.worst.catKey === 'wrinkles' && vm.group === '25+组'
-        && vm.essence.name === '35+支撑精华' && vm.essence.triggers.join('、') === '最差项约束',
-        '均分 9.05 / 最差皱纹 / 25+组 / 精华 35+支撑（触发：最差项约束，三重判据取最强档）');
+        && vm.essence.name === '25+束紧精华' && vm.essence.triggers.join('、') === '肤龄基础分组',
+        '评分不变，原始肌龄27推荐25+，不再因最差皱纹升档');
       ok(vm.skinAgeRaw === 27 && vm.skinAgeDisplay === 27 && vm.ageClampNote === '', '肤龄 27（实际30）未触发钳制、原值不改写');
       ok(vm.cats.contour.text === '下颌线角度左104.54°/右103.75°（标准区间116–120°，双侧均偏离标准，提示下颌缘支撑力下降）；眼袋1级、黑眼圈1级；中面部支撑与轮廓清晰度处于预警水平。建议加强轮廓紧致护理：重点针对下颌线与下颌缘区域使用紧致类精华，配合由下向上的提拉按摩手法，同时加强眼周支撑护理，将紧致防线前置。',
         '轮廓文案逐字 = 真值表（黑眼圈取 dark_circle_severity=1，非 dark_circle=3）');
@@ -1153,8 +1186,8 @@ if (typeof require === 'function' && typeof module !== 'undefined' && require.ma
         '敏感文案逐字 = 真值表');
       ok(vm.cats.acne.text === '检出9处皮损（闭口粉刺6处、丘疹3处）。建议尽快进行针对性抗痘护理，避免挤压与刺激，必要时寻求皮肤科专业处理。',
         '痘痘文案逐字 = 真值表（无 heavy_note）');
-      ok(vm.summary === '【综合】实际年龄30岁（接口测得肌肤年龄27岁，差3岁），产品组别25+组；六分类S均分9.05、最差分类皱纹 S=7.69。按三重判据（触发：最差项约束）推荐35+支撑精华：优先改善皱纹（P=23.0）与光滑（P=57.2），其余维度维持现有护理节奏。整体肤质混合性，T区出油面积57%、左右脸颊20%/2%。',
-        '综合文案逐字 = 真值表 AW 列');
+      ok(vm.summary === '【综合】实际年龄30岁（接口测得肌肤年龄27岁，差3岁），产品组别25+组；六分类S均分9.05、最差分类皱纹 S=7.69。按肤龄及边界规则（触发：肤龄基础分组）推荐25+束紧精华：优先改善皱纹（P=23.0）与光滑（P=57.2），其余维度维持现有护理节奏。整体肤质混合性，T区出油面积57%、左右脸颊20%/2%。',
+        '综合文案仅产品判据按2026-10-12规则替换，其余逐字不变');
       ok(vm.lifestyle === '保持侧卧改仰卧的睡姿习惯，减少枕巾摩擦；控制屏幕时间，缓解眼周疲劳', '生活方式 = 最差分类皱纹条');
     }());
 

h5/lib/accounts.js

@@ -157,5 +157,13 @@ module.exports = function createAccounts(root, env = process.env) {
       .concat(user.reports.filter(r => r.id !== snapshot.id)).slice(0, 5);
     save();
   }
-  return { route, session, storeReport, sameOrigin, wechatAvailable: available };
+  // Internal admin use only: reuse historical copy, never regenerate it.
+  // Not a public route; validate the stored owner and report ID before reading.
+  function reportPresentation(ownerId, id) {
+    const user = db.users.find(u => u.id === ownerId);
+    if (!user || !/^R[\w-]+$/.test(id)) return null;
+    try { return JSON.parse(fs.readFileSync(path.join(snapshots, user.id, id + '.json'), 'utf8')).presentation || null; }
+    catch (_) { return null; }
+  }
+  return { route, session, storeReport, reportPresentation, sameOrigin, wechatAvailable: available };
 };

h5/server.js

@@ -257,6 +257,9 @@ function fillSuccess(rec, fpJson) {
   rec.detect_status = '成功';
   rec.fail_stage = null; rec.fail_code = null; rec.fail_msg = null;
   rec.skin_age = vm.skinAgeRaw;
+  rec.skin_age_display = vm.skinAgeDisplay;
+  rec.recommendation = vm.recommendation;
+  rec.presentation = ReportUI.build(vm, fpJson.result);
   rec.total_score = vm.total;
   rec.jawline_angle_l = jawlineAngle(fpJson.left_side_result, 'left');
   rec.jawline_angle_r = jawlineAngle(fpJson.right_side_result, 'right');
@@ -433,7 +436,7 @@ async function handleAnalyze(req, res) {
     let reportSaved = false, presentation = null, snapshotVm = null;
     try {
       snapshotVm = SkinScore.buildViewModel(result, payload.side, scoreProfile(rec));
-      presentation = ReportUI.build(snapshotVm, result);
+      presentation = rec.presentation || ReportUI.build(snapshotVm, result);
       accounts.storeReport(owner.id, { id: recordId, created_at: rec.created_at,
         version: '2026-10-09', profile, vm: snapshotVm, presentation, payload,
         photos: [front, left, right].map(b64 => b64 ? toDataUrl(b64) : null) });
@@ -482,6 +485,12 @@ function photoUrls(rec) {
 }
 
 /* 列表用的字段子集(不含 raw_json 与文案全文) */
+function displaySkinAge(rec) {
+  if (Object.prototype.hasOwnProperty.call(rec, 'skin_age_display')) return rec.skin_age_display;
+  // 旧记录没有展示值，沿用生成时的规则，不重算推荐或改写文件。
+  if (rec.skin_age == null || rec.real_age == null) return rec.skin_age;
+  return Math.min(Math.max(rec.skin_age, rec.real_age - 3), rec.real_age + 3);
+}
 function listItem(rec) {
   return {
     record_id: rec.record_id,
@@ -490,6 +499,7 @@ function listItem(rec) {
     shoot_status: rec.shoot_status, upload_status: rec.upload_status, detect_status: rec.detect_status,
     fail_code: rec.fail_code, fail_msg: rec.fail_msg,
     total_score: rec.total_score, skin_age: rec.skin_age,
+    skin_age_display: displaySkinAge(rec), recommendation: rec.recommendation || null,
     dims: rec.dims, group: rec.group,
     essence: rec.essence ? rec.essence.name : null,
     cost_ms: rec.cost_ms,
@@ -576,7 +586,8 @@ function handleRecordDetail(req, res, id) {
     .map(r => ({ record_id: r.record_id, detect_status: r.detect_status, created_at: r.created_at }));
   sendJson(res, 200, {
     ok: true,
-    record: rec,
+    record: Object.assign({}, rec, { skin_age_display: displaySkinAge(rec),
+      presentation: rec.presentation || accounts.reportPresentation(rec.owner_id, rec.record_id) }),
     raw_json: rawJson,
     photos: photoUrls(rec),
     versions: {
@@ -733,21 +744,24 @@ function handleExport(req, res, url) {
     .sort((a, b) => (a.created_at < b.created_at ? 1 : -1));
   const dimP = (r, d) => (r.dims && r.dims[d]) ? r.dims[d].P : null;
   const dimS = (r, d) => (r.dims && r.dims[d]) ? r.dims[d].S : null;
-  const dimLv = (r, d) => (r.dims && r.dims[d]) ? r.dims[d].level : null;
+  const dimLv = (r, d) => (r.dims && r.dims[d]) ? ReportUI.levelMap[r.dims[d].level] : null;
   const header = ['记录ID', '创建时间', '年龄', '性别', '地区', '日常肤质', '拍摄状态', '上传状态', '检测状态',
-    '失败环节', '失败码', '失败信息', '总分', '肌肤年龄',
-    '紧塑力P', '紧塑力S', '紧塑力档', '活力感P', '活力感S', '活力感档', '稳定力P', '稳定力S', '稳定力档',
-    '均分S', '产品组别', '推荐精华', '触发判据', '耗时ms'];
+    '失败环节', '失败码', '失败信息', '总分', '原始肌龄（推荐依据）', '前端展示肌龄',
+    '紧塑力P', '紧塑力S', '紧塑力档', '透亮感P', '透亮感S', '透亮感档', '稳韧力P', '稳韧力S', '稳韧力档',
+    '均分S', '产品组别', '推荐精华', '触发判据', '耗时ms', '推荐规则版本', '推荐原因'];
   const lines = [header.join(',')];
   for (const r of filtered) {
     lines.push([
       r.record_id, r.created_at, r.real_age, r.gender, r.region, r.daily_skin_type,
       r.shoot_status, r.upload_status, r.detect_status,
-      r.fail_stage, r.fail_code, r.fail_msg, r.total_score, r.skin_age,
+      r.fail_stage, r.fail_code, r.fail_msg, r.total_score, r.skin_age, displaySkinAge(r),
       dimP(r, 'firm'), dimS(r, 'firm'), dimLv(r, 'firm'),
       dimP(r, 'vita'), dimS(r, 'vita'), dimLv(r, 'vita'),
       dimP(r, 'stab'), dimS(r, 'stab'), dimLv(r, 'stab'),
-      r.avg_s, r.group, r.essence ? r.essence.name : null, r.trigger, r.cost_ms
+      r.avg_s, r.group, r.essence ? ((ReportUI.products.find(p => p.match === r.essence.name) || {}).title || r.essence.name) : null,
+      r.trigger, r.cost_ms,
+      r.recommendation ? r.recommendation.version : '旧版规则（历史记录）',
+      r.recommendation ? r.recommendation.reason : null
     ].map(csvCell).join(','));
   }
   const fromTag = (q.from || 'all').replace(/[^\d-]/g, '');

h5/test/accounts.js

@@ -37,6 +37,10 @@ const ReportUI = require('../js/report-ui');
       payload: { result: { ...SkinScore.MOCK_RESULT, face_maps: { brown_area: 'data:image/png;base64,saved-layer' } }, side: SkinScore.MOCK_SIDE } };
     accounts.storeReport(a.data.user.id, snapshot);
     assert.deepEqual((await request('/api/reports/Rtest0', null, a.cookie)).data.report, JSON.parse(JSON.stringify(snapshot)));
+    assert.deepEqual(accounts.reportPresentation(a.data.user.id, 'Rtest0'), JSON.parse(JSON.stringify(presentation)));
+    assert.equal(accounts.reportPresentation(b.data.user.id, 'Rtest0'), null);
+    assert.equal(accounts.reportPresentation(a.data.user.id, '../identities'), null);
+    assert.equal(accounts.reportPresentation('../reports', 'Rtest0'), null);
     assert.equal((await request('/api/reports/Rtest0', null, b.cookie)).status, 404);
     assert.equal((await request('/api/reports', null, b.cookie)).data.total, 0);
     for (let i = 1; i <= 5; i++) accounts.storeReport(a.data.user.id, { ...snapshot, id: 'Rtest' + i });

h5/test/recommendation.js

@@ -0,0 +1,67 @@
+'use strict';
+const assert = require('node:assert/strict');
+const SS = require('../js/skin-score.js');
+const UI = require('../js/report-ui.js');
+let count = 0;
+function check(age, firm, vita, stab, tier, adjustment = 'none') {
+  const r = SS.recommendProduct(age, { firm, vita, stab });
+  assert.equal(r.tier, tier, `age=${age}; scores=${firm}/${vita}/${stab}`);
+  assert.equal(r.adjustment, adjustment);
+  assert.equal(r.version, '2026-10-12-skin-age-v1');
+  if (tier) assert(Math.abs(r.tier - r.baseTier) <= 1);
+  count++;
+}
+// Explicit decision-table expectations, including fractional ages and inclusive endpoints.
+for (const [age, normal, low, high] of [
+  [24, 1, 1, 1], [31.99, 1, 1, 1], [32, 1, 2, 1], [34.99, 1, 2, 1],
+  [35, 2, 2, 1], [36, 2, 2, 1], [37, 2, 2, 1], [38, 2, 2, 1], [38.01, 2, 2, 2],
+  [46.99, 2, 2, 2], [47, 2, 3, 2], [49.99, 2, 3, 2],
+  [50, 3, 3, 2], [51, 3, 3, 2], [52, 3, 3, 2], [53, 3, 3, 2], [53.01, 3, 3, 3], [80, 3, 3, 3]
+]) {
+  check(age, 8, 8, 8, normal);
+  check(age, 7.99, 10, 10, low, low > normal ? 'up' : 'none');
+  check(age, 9, 9, 7, high, high < normal ? 'down' : 'none');
+  check(age, 9, 7, 9, high, high < normal ? 'down' : 'none');
+}
+// Thresholds are compared before display rounding, not to 1- or 2-decimal scores.
+for (const age of [32, 34.99, 47, 49.99]) {
+  const base = age < 35 ? 1 : 2;
+  check(age, 7.99999, 10, 10, base + 1, 'up');
+  check(age, 8, 10, 10, base);
+  check(age, null, 10, 10, base);
+}
+for (const age of [35, 38, 50, 53]) {
+  const base = age < 50 ? 2 : 3;
+  check(age, 8.99999, 10, 10, base);
+  check(age, 9, 8.99999, 8.99999, base);
+  check(age, 9, null, null, base);
+  check(age, null, 10, 10, base);
+  check(age, 9, null, 9, base - 1, 'down');
+  check(age, 9, 9, null, base - 1, 'down');
+}
+check(37, 9, 7, 9, 1, 'down'); // Only dullness: the third dimension may be arbitrarily low.
+check(35, 8.99, 10, 10, 2); // Firmness is necessary, other two alone do not downgrade.
+check(24, null, null, null, 1);
+check(40, null, null, null, 2);
+for (const invalid of [null, undefined, NaN, Infinity, -Infinity, 0, -5, '37', {}]) {
+  check(invalid, 10, 10, 10, null, 'unavailable');
+}
+const raw = JSON.parse(JSON.stringify(SS.MOCK_RESULT));
+raw.skin_age = { value: 24 };
+for (const realAge of [24, 40, 60, undefined]) {
+  const vm = SS.buildViewModel(raw, SS.MOCK_SIDE, { realAge });
+  assert.equal(vm.essence.tier, 1); // Weak wrinkles never upgrade under32.
+  assert.equal(vm.recommendation.skinAge, 24);
+  assert(Math.abs(vm.recommendation.scores.firm - 7.69) < 1e-12);
+  assert.equal(UI.build(vm, raw).product.tier, 1);
+  if (realAge === 60) assert.equal(vm.skinAgeDisplay, 57); // Display isn't the recommendation input.
+  count++;
+}
+delete raw.skin_age;
+const missing = SS.buildViewModel(raw, SS.MOCK_SIDE, { realAge: 60 });
+assert.equal(missing.group, null);
+assert.equal(missing.essence.tier, null);
+assert.equal(UI.build(missing, raw).product, null);
+assert(missing.recommendation.reason.includes('未获取有效原始肌龄'));
+count++;
+console.log(`PASS: ${count} recommendation cases; boundary/threshold/missing/raw-age/actual-age isolation`);

h5/test/regression.js

@@ -8,7 +8,8 @@
  *  - 6 分类文案(L/P/T/X/AB/AF)、综合文案(AW)、生活方式(AX) 逐字一致
  *  - 浮点容差 0.05；文案逐字无容差
  *
- * 引擎与真值不一致时：不改 h5/js/skin-score.js（审计代理负责），差异登记在 KNOWN_DIFFS
+ * 产品推荐按2026-10-12用户确认规则覆盖旧Excel期望；评分真值和原始fixture不改写。
+ * 非授权的新差异登记在 KNOWN_DIFFS
  * 并输出明细；未登记的新差异 → 退出码 1。登记项若转为通过（引擎已修复）会提示清理。
  */
 'use strict';
@@ -20,6 +21,12 @@ var SkinScore = require(path.join(__dirname, '..', 'js', 'skin-score.js'));
 /* ─── 已知差异登记：键 = 案例.列，值 = 差异原因（逐条说明）。当前为空：V2.2 审计后
    引擎与两组真值全部一致；今后出现引擎/真值不一致时在此登记，未登记的新差异 → 退出码 1 ─── */
 var KNOWN_DIFFS = {};
+var RECOMMENDATION_EXPECT = {
+  case1: { essence: '25+束紧精华', oldSummary: '按三重判据（触发：最差项约束）推荐35+支撑精华',
+    newSummary: '按肤龄及边界规则（触发：肤龄基础分组）推荐25+束紧精华' },
+  case2: { essence: '50+重塑精华', oldSummary: '按三重判据（触发：年龄组档）推荐50+重塑精华',
+    newSummary: '按肤龄及边界规则（触发：肤龄基础分组）推荐50+重塑精华' }
+};
 
 /* ─── V2.3 §8.8 护理建议成品真值(五段,逐字) ─── */
 var V23_SKIN = { case1: '混合偏干肌肤', case2: '干性肌肤' };
@@ -28,7 +35,7 @@ var V23_EXPECT = {
     { tag: '肤质', text: '你的日常肤质为混合偏干肌肤：T区状态尚可，但两颊及面部外侧偏干，易出现紧绷与细纹。建议分区护理：两颊侧重保湿滋养，选用滋润型乳液/面霜，T区轻量保湿即可，避免全脸统一用厚重面霜。干燥会放大纹理粗糙感，建议先将两颊水合度提上来再做角质调理。' },
     { tag: '皱纹', text: '加强纹路分级管理，重点针对眼周、法令纹等动态纹高发区域，使用视黄醇/胜肽类淡纹精华并配合点按手法，遏制静态纹转化。' },
     { tag: '光滑', text: '加强细腻度管理，避免过度清洁破坏角质屏障，重点关注毛孔显著区域，配合低浓度酸类（如水杨酸）周期护理与柔润精华。' },
-    { tag: '精华', text: '建议搭配使用35+支撑精华：专为轻熟肌打造，提升弹韧表现、维持面部轮廓清晰度，改善疲态并延缓衰老速度。' },
+    { tag: '精华', text: '建议搭配使用25+束紧精华：针对年轻肌肤活力减弱问题，激活肌肤动力、改善暗沉，提升弹润紧致表现，守护年轻光泽并延缓初始衰老迹象。' },
     { tag: '生活方式', text: '保持侧卧改仰卧的睡姿习惯，减少枕巾摩擦；控制屏幕时间，缓解眼周疲劳' }
   ],
   case2: [
@@ -120,9 +127,10 @@ function main() {
     }
     // 组别/精华/均分/综合文案/生活方式
     okStr(tag + '.AT 产品组别', String(vm.group), tr.group);
-    okStr(tag + '.AU 精华产品线', vm.essence.name, tr.essence);
+    okStr(tag + '.AU 精华产品线（肤龄新版）', vm.essence.name, RECOMMENDATION_EXPECT[tag].essence);
     okNum(tag + '.AV 六分类S均分', vm.avgS, tr.avgS);
-    okStr(tag + '.AW 综合文案', vm.summary, tr.summary, tag + '.AW');
+    okStr(tag + '.AW 综合文案', vm.summary, tr.summary.replace(RECOMMENDATION_EXPECT[tag].oldSummary,
+      RECOMMENDATION_EXPECT[tag].newSummary), tag + '.AW');
     okStr(tag + '.AX 生活方式', vm.lifestyle, tr.lifestyle);
 
     // V2.3 §8 护理建议五段(真值:§8.8 样例;daily_skin_type 按样例取,与 summary 口径的肤质参数不同)

h5/test/report-persistence.mjs

@@ -8,6 +8,9 @@ import http from 'node:http';
 import { spawn, execSync } from 'node:child_process';
 import { once } from 'node:events';
 import { fileURLToPath, pathToFileURL } from 'node:url';
+import { createRequire } from 'node:module';
+const require = createRequire(import.meta.url);
+const ReportUI = require('../js/report-ui.js');
 const { chromium } = await import(pathToFileURL(execSync('npm root -g', { encoding: 'utf8' }).trim() + '/playwright/index.mjs').href);
 const h5 = fileURLToPath(new URL('../', import.meta.url));
 const dataRoot = fs.mkdtempSync(path.join(os.tmpdir(), 'juzi-report-test-'));
@@ -40,15 +43,68 @@ try {
   await page.click('[data-action="login-browser"]');
   await page.waitForSelector('#s-splash.active');
   const api = page.context().request;
-  const result = await (await api.post(base + 'api/skin-analyze', { data: { front_image: fixture.photos[0], left_image: fixture.photos[0], right_image: fixture.photos[0], profile: { real_age: 28, daily_skin_type: '干性肌肤' } } })).json();
+  const result = await (await api.post(base + 'api/skin-analyze', { data: { front_image: fixture.photos[0], left_image: fixture.photos[0], right_image: fixture.photos[0], profile: { real_age: 60, daily_skin_type: '干性肌肤' } } })).json();
   assert.equal(result.ok, true);
   assert.equal(result.report_saved, true);
   assert(result.vm && result.presentation);
+  assert(result.vm.skinAgeRaw < 32);
+  assert.equal(result.vm.skinAgeDisplay, 57);
+  assert.equal(result.vm.essence.tier, 1); // Raw age, not actual60 or display57.
+  assert.equal(result.presentation.product.tier, 1);
+  const adminRecord = (await (await api.get(base + 'api/admin/records/' + result.record_id)).json()).record;
+  assert.deepEqual(adminRecord.presentation, result.presentation);
+  assert.deepEqual(adminRecord.recommendation, result.vm.recommendation);
+  assert.equal(adminRecord.skin_age, result.vm.skinAgeRaw);
+  assert.equal(adminRecord.skin_age_display, result.vm.skinAgeDisplay);
+  for (const key of ['firm', 'vita', 'stab']) assert.equal(adminRecord.dims[key].S, result.vm.dims[key].S);
+  const list = await (await api.get(base + 'api/admin/records')).json();
+  assert.equal(list.items[0].skin_age, result.vm.skinAgeRaw);
+  assert.equal(list.items[0].skin_age_display, 57);
+  assert.deepEqual(list.items[0].recommendation, result.vm.recommendation);
+  const csv = await (await api.get(base + 'api/admin/export')).text();
+  assert(csv.includes('原始肌龄（推荐依据）,前端展示肌龄'));
+  assert(csv.includes(result.vm.recommendation.version));
   const snapshot = await (await api.get(base + 'api/reports/' + result.record_id)).json();
   assert.equal(snapshot.report.photos.length, 3);
   assert.deepEqual(snapshot.report.presentation, result.presentation);
   assert.deepEqual(snapshot.report.payload.result.face_maps, result.result.face_maps);
   assert.equal(snapshot.report.photos[0], fixture.photos[0]);
+  const desktop = await browser.newPage({ viewport: { width: 1600, height: 1000 } });
+  desktop.on('pageerror', e => errors.push(e.message));
+  await desktop.goto(base + 'admin.html');
+  const row = desktop.locator('#tbl-records tbody tr').first();
+  await row.waitFor();
+  assert((await row.textContent()).includes('原始 ' + result.vm.skinAgeRaw + '岁'));
+  assert((await row.textContent()).includes('展示 57岁'));
+  assert((await row.textContent()).includes("1'me 主场精华 I（25+）"));
+  for (const key of ['firm', 'vita', 'stab']) {
+    assert((await row.textContent()).includes(ReportUI.dimNames[key] + ' ' + result.vm.dims[key].S.toFixed(1) + ' ' + ReportUI.levelMap[result.vm.dims[key].level]));
+  }
+  assert.deepEqual(await desktop.locator('#f-dimlevel option').allTextContents(), ['不限档位', '改善', '良好', '优秀']);
+  if (process.env.H5_ARTIFACT_DIR) {
+    fs.mkdirSync(process.env.H5_ARTIFACT_DIR, { recursive: true });
+    await desktop.screenshot({ path: path.join(process.env.H5_ARTIFACT_DIR, 'admin-list.png'), fullPage: true });
+  }
+  await desktop.selectOption('#f-dim', 'firm');
+  await desktop.selectOption('#f-dimlevel', adminRecord.dims.firm.level);
+  const filtered = desktop.waitForResponse(r => r.url().includes('/api/admin/records?'));
+  await desktop.click('#btn-search');
+  assert.equal((await (await filtered).json()).total, 1);
+  await row.click();
+  await desktop.waitForFunction(() => document.querySelector('#d-body').textContent.includes('推荐依据'));
+  const detailText = await desktop.locator('#d-body').textContent();
+  assert(detailText.includes(result.vm.recommendation.reason));
+  assert(detailText.includes('前端展示肌龄 57岁'));
+  for (const view of result.presentation.views) assert(detailText.includes(view.text));
+  for (const care of result.presentation.care) assert(detailText.includes(care.text));
+  if (process.env.H5_ARTIFACT_DIR) {
+    await desktop.locator('.d-sec').filter({ hasText: '使用原始肌龄' }).scrollIntoViewIfNeeded();
+    await desktop.screenshot({ path: path.join(process.env.H5_ARTIFACT_DIR, 'admin-recommendation.png') });
+  }
+  await desktop.click('#btn-close');
+  await desktop.click('[data-tab="stats"]');
+  await desktop.waitForFunction(() => document.querySelector('#dim-dist').textContent.includes('透亮感'));
+  assert(!(await desktop.locator('#dim-dist').textContent()).includes('预警'));
   // A separate browser cannot access this user's report, even knowing its ID.
   const other = await browser.newContext();
   assert.equal((await other.request.get(base + 'api/reports/' + result.record_id)).status(), 401);
@@ -78,12 +134,55 @@ try {
   await page.waitForFunction(() => [...document.querySelectorAll('.ditem-img')].every(i => !i.hidden && i.naturalWidth));
   assert.deepEqual(await page.locator('.ditem-img').evaluateAll(images => images.map(i => i.src)), imagesBefore);
   assert.deepEqual(await page.locator('.ditem-para').allTextContents(), copyBefore);
+  assert.equal(await page.locator('#product-title').textContent(), result.presentation.product.title);
   await page.goBack();
   await page.waitForSelector('#s-history.active');
   await page.goBack();
   await page.waitForSelector('#s-splash.active');
+  // Simulate an old, immutable report whose original product was II, even though
+  // today's rule would choose I. Admin must reuse its saved frontend copy, too.
+  const recordFile = path.join(dataRoot, 'records.jsonl');
+  const record = JSON.parse(fs.readFileSync(recordFile, 'utf8').trim());
+  const snapshotFile = path.join(dataRoot, 'accounts/reports', record.owner_id, result.record_id + '.json');
+  const legacy = JSON.parse(fs.readFileSync(snapshotFile, 'utf8'));
+  legacy.presentation.product = ReportUI.products[1];
+  legacy.presentation.views[0].text = '历史报告原文，不应按新文案重新计算。';
+  legacy.vm.essence = { name: '35+支撑精华', text: '旧版推荐', triggers: ['最差项约束'] };
+  delete legacy.vm.recommendation;
+  fs.writeFileSync(snapshotFile, JSON.stringify(legacy));
+  record.essence = legacy.vm.essence;
+  delete record.presentation;
+  delete record.recommendation;
+  delete record.skin_age_display;
+  fs.writeFileSync(recordFile, JSON.stringify(record) + '\n');
+  const storedBytes = fs.readFileSync(snapshotFile, 'utf8');
+  const recordBytes = fs.readFileSync(recordFile, 'utf8');
+  await page.click('[data-action="history"]');
+  await page.click('[data-report-id]');
+  await page.waitForSelector('#s-report.active');
+  assert.equal(await page.locator('#product-title').textContent(), ReportUI.products[1].title);
+  assert((await page.locator('.ditem-para').allTextContents()).includes(legacy.presentation.views[0].text));
+  await desktop.goto(base + 'admin.html');
+  await desktop.locator('#tbl-records tbody tr').first().click();
+  await desktop.waitForFunction(() => document.querySelector('#d-body').textContent.includes('历史报告原文'));
+  assert((await desktop.locator('#d-body').textContent()).includes('旧版规则（历史记录）'));
+  assert.equal(fs.readFileSync(snapshotFile, 'utf8'), storedBytes);
+  assert.equal(fs.readFileSync(recordFile, 'utf8'), recordBytes);
+  // Missing skin age must not leave the previous product visible in the H5.
+  delete legacy.payload.result.skin_age;
+  const SkinScore = require('../js/skin-score.js');
+  legacy.vm = SkinScore.buildViewModel(legacy.payload.result, legacy.payload.side, { realAge: 60 });
+  legacy.presentation = ReportUI.build(legacy.vm, legacy.payload.result);
+  fs.writeFileSync(snapshotFile, JSON.stringify(legacy));
+  await page.reload();
+  await page.waitForSelector('#s-splash.active');
+  await page.click('[data-action="history"]');
+  await page.click('[data-report-id]');
+  await page.waitForSelector('#s-report.active');
+  assert.equal(await page.locator('.card-product').isVisible(), false);
+  assert.equal(await page.locator('#product-unavailable').isVisible(), true);
   assert.deepEqual(errors, []);
-  console.log('PASS: authenticated analyze → full snapshot → backend history → reload → seven identical images/copy; owner isolation, private files, browser back');
+  console.log('PASS: raw-age recommendation; admin/H5 scores, ages, product, seven-view copy, care, CSV, filters; saved/reloaded and legacy immutable snapshots; missing-age UI; owner isolation/private files/back');
 } finally {
   if (browser) await browser.close();
   child.kill('SIGTERM');