feat(CaptureService): 改进报告解析逻辑并添加多语言键支持

- 修改应用描述从"个人健康影像档案"到"个人健康随记"
- 添加对多种JSON键名的支持,包括中文键名(如"指标"、"项目"、"结果"等)
- 实现指标状态智能推断功能,可根据数值和参考范围自动判断高低状态
- 支持多种状态标识符,包括箭头符号(↑↓)和中英文状态词
- 增加对不同参考范围格式的解析支持(如"< 3.40"、"208 - 428"等)
- 添加相关单元测试验证中文键名和状态推断功能
```
This commit is contained in:
link2026
2026-06-06 12:53:52 +08:00
parent 77697e1600
commit 675c33bea1
6 changed files with 540 additions and 20 deletions

290
AGENTS.md Normal file
View File

@@ -0,0 +1,290 @@
# 康康 —— 工程前提
> 这是一个 6 周决赛 demo 项目。今天是 2026-05-25,处于 W1末/W2初。
> 任何 IDE/Codex 会话开始干活前,先读这份文件。
---
## 1. 产品定位
- **名字**:康康(对内代号 Kangkang)
- **形态**:iOS 原生 App,SwiftUI + SwiftData
- **核心卖点**:**100% 本地推理**的个人健康影像档案 + 大白话解读 + 本地 RAG 问答
- **目标用户**:不愿把体检/化验报告交给云端的普通人
- **明确不做**:医疗诊断、剂量推荐、急诊判断、医生预约、社交、广告、内购、数据上云、账号系统
---
## 2. 技术栈 / 选型(已锁定,不要再讨论)
| 项 | 选型 | 备注 |
|---|---|---|
| UI | SwiftUI | iOS 17+,用 `@Observable` / `@Model` |
| 持久化 | SwiftData | 见 §5 数据模型 |
| 图表 | Swift Charts | iOS 16+ 原生 |
| **AI 运行时** | **MLX Swift (Apple 官方)** | 不要建议 Core ML / llama.cpp / Ollama |
| LLM | Qwen3-1.7B 4bit (HF: `mlx-community/Qwen3-1.7B-4bit`) | ~1.0GB,负责文本生成、关键词抽取、趋势解读 |
| VL | Qwen2.5-VL-3B-Instruct 4bit (HF: `mlx-community/Qwen2.5-VL-3B-Instruct-4bit`) | ~2.0GB,负责拍照→结构化指标 |
| 文档扫描 | VisionKit `VNDocumentCameraView` | 不要自己写透视校正 |
| Face ID | LocalAuthentication | |
| Live Activity | ActivityKit + WidgetExtension | demo 杀手锏,真机才能测 |
**不引入**:任何云服务 SDK、任何 embedding 模型(RAG 用结构化检索,不用语义)、任何账号系统、任何分析 SDK。
---
## 3. AI 链路核心规则
### 3.1 模块边界(强制)
```
UI → CaptureService / AskService / TrendService → AIRuntime → MLX
Persistence
```
- **UI 永远不直接调 `AIRuntime`**。所有 AI 调用必须经过 `*Service` 层,这样 UI 可以注入 mock、可以预览。
- **`AIRuntime``actor` 单例,串行化**。同一时刻只允许一个推理任务,MLX 共享显存,并发会 OOM。CaptureService 拍照时如果 AskService 正在流式生成,要在队列里排队。
- **`*Service` 不直接读写 SwiftData 主上下文**。要么传入 `ModelContext`,要么走 ServiceLocator,方便测试。
### 3.2 VL pipeline(拍一张 = 一条流程)
**重要**:快拍(1.x) 和 报告归档(2.x) 已经合并成统一 `CaptureService`,UI 不再有 A1-A3 和 B1-B4 两条独立路径。流程:
```
拍照 → 写 Vault(加密目录) → VL 推理(要求输出 JSON,含 kind=single|report)
→ 解析容错(失败回退到手动录入,不卡死)
→ 单项走 A2ConfirmView,整份走 B3MetaView
→ 保存到 Indicator/Report + 关联 Asset
```
VL prompt 必须:
- 明确要求"只输出 JSON,不要解释"
- 带 2 个 few-shot 示例(单项 + 多项)
- 异常状态由 VL 模型基于参考范围直接判断,不要再二次调用 LLM
### 3.3 RAG(结构化检索,不做 embedding)
**两段式调用**:
1. 用 Qwen3-1.7B 抽取意图 + 关键词,输出 JSON `{indicators, time_range, intent}`,~50 token,<1s
2. SwiftData 按关键词检索 ≤ 10 条记录,拼 `ChatRAG` prompt,流式生成回答
**第 1 步失败时**回退到"近 30 天全表扫描",不卡死。
**引用回链**:回答中 `[1][2]` 后处理为可点击 Pill,点击跳源记录详情。
### 3.4 Live Activity
- VL 推理 / RAG 生成开始时启动 Activity
- 每 0.5s 通过 `AIRuntime.lastDecodeRate` 推送 tok/s
- 推理完成保留 2s 显示"已完成 · 0.8s"再 dismiss
- **只能真机测**,模拟器不显示。W5 末预留时间。
---
## 4. 模型分发
- 模型放 `Application Support/Models/`,首启动用 `URLSession.downloadTask` 拉,带断点续传 + 进度条
- 总体积 ~4GB(LLM ~1.0GB + VL ~3.1GB),WiFi 提示必须有
- App 在模型未就绪时**仍可启动**,但所有 AI 入口显示"模型未就绪,前往下载"
- `ModelStore` 必须提供**旁路接口**:允许把模型预拷进沙盒(demo 现场重装时用)
---
## 5. 数据模型(SwiftData)
**当前 schema(2026-05-26)**:7 个 @Model
```swift
@Model class Indicator {
name, value, unit, range, statusRaw, note, capturedAt,
report: Report?, asset: Asset?,
pinned: Bool, // true,Trends
seriesKey: String? // "bp.systolic" / "glucose.fasting" / ... key
}
@Model class Report { title, typeRaw, reportDate, institution, note, summary, pageCount, createdAt,
indicators: [Indicator] cascade,
assets: [Asset] cascade }
@Model class DiaryEntry { content, createdAt, tags: [String] }
@Model class Symptom { name, startedAt, endedAt?, note?, severity 1-5, tags, createdAt }
@Model class Asset { relativePath, mimeType, bytes, createdAt }
@Model class ChatTurn { question, answer, referencedIndicatorIDs, referencedReportIDs, createdAt, decodeRate }
@Model class UserProfile { // App (UserProfileStore.loadOrCreate)
birthYear?, biologicalSexRaw, heightCM?, bloodTypeRaw,
allergies, chronicConditions, familyHistory, currentMedications,
updatedAt
}
```
**原图存储**: `Asset` 只存元数据 + 相对路径,真实 JPEG 落在 `Application Support/Vault/`,目录用 `.completeFileProtection`(iOS 硬件级加密,不要自己造 AES 轮子)。
---
## 6. 安全 / 隐私(已收敛 — 不要扩展)
| 做 | 不做 |
|---|---|
| `Application Support/Vault/` 全目录 `.completeFileProtection` | 自实现 AES 加密 |
| SwiftData store 文件 `.completeFileProtection` | |
| Face ID 启动锁(可选开关,默认关) | |
| 永久删除(SwiftData 硬删 + Asset 文件 unlink) | |
| 离线运行(自然结果,不用单独做) | |
| | 截屏黑屏防护(iOS 没有官方 API,不做) |
| | 加密 ZIP 导出 |
唯一的"导出"是 **9.4 分享文字摘要**(只分享解读文本,不带原图)。
---
## 7. 信息架构
```
TabBar: [主页] [记录] [+ 新建] [趋势] [我的]
│ │ │ │ │
│ │ │ │ └─ 个人资料 / 模型管理 / Face ID / 关于
│ │ │ └─ 折线图 + AI 一句话解读
│ │ └─ Sheet: 拍一张 / 指标记录 / 报告归档 / 写日记 / 症状
│ └─ ArchiveListView(时间线 + 分类 chip + 年/月分组)
└─ 问候 + 今日摘要 + 进行中症状 + 最近时间线
```
- TabBar **5 槽**:左 2 个内容 Tab + 中间 + 号 + 右 2 个 Tab
- "+ 新建" 是 sheet 不是 Tab
- AI 问答以 Modal Sheet 形式出现,**不占 Tab**
- 「指标记录」sheet 顶部 LazyVGrid 是 8 个 MonitorMetric 长期监测预设(进趋势),
下方 horizontal scroll 是化验项快捷预设(不进趋势),不选预设走自由输入
- 「我的 · 个人资料」是 NavigationLink push 的 Form 编辑页
### 7.1 档案库 C1 / C2 导航(看的一半)
录入流程(拍照→VL→编辑→存)只是"录的一半"。**"看的一半"由 C1 列表 + C2 详情承担**——这是 demo 的核心看点之一,不能砍。
```
首页 "我的报告档案" 卡 ──push──► C1 ArchiveListView
│ 分类 chip:全部/体检/化验/影像/处方
│ 按 reportDate 年份分组,卡片显示异常 chip
└──push──► C2 ReportDetailView
├─ Tab "原图":TabView(.page) 翻页 + 长按保存
├─ Tab "解读":数字摘要(总/高/低/正常)
│ + AI 整体摘要
│ + 对比上次(同类型上一份 Report diff)
└─ Tab "指标":Indicator 列表,异常优先
C2 底部两个动作:
├─ "关联到趋势" ──► 把本报告内未 pinned 的 Indicator 批量 pinned = true,Trends 默认展示
└─ "重新解读" ──► CaptureService.reanalyze(report:),重跑 VL 覆盖 summary/indicators
其他进入 C2 的入口:
• ChatTurn 引用 Pill 点击(referencedReportIDs)
• 趋势页数据点 tap → 跳到该点来源 Report 的 C2
• HomeView 时间线点报告类条目
```
### 7.2 对比上次("对比上次"=报告对比,已加回)
C2 解读 Tab 底部显示一段 diff 文本,**由 `ReportCompareService` 计算,不再调 LLM**:
- 找出"同 `typeRaw` 的上一份 Report"(`reportDate < current AND ORDER BY DESC LIMIT 1`)
- 同名 `Indicator` 配对,数值 diff:`Δ` 绝对值 + 百分比 + 升/降箭头
- 标红:跨越参考范围边界(原本正常→偏高,或反过来)
- 文案模板拼装,不走 LLM,响应即时
- 若无上一份,该区块隐藏
---
## 8. 现有代码状态(2026-05-25)
```
康康/
├── App/KangkangApp.swift ✅ SwiftData container 已建
├── RootView.swift ✅ 3 Tab + RecordSheet 已建
├── Models/Models.swift ✅ Indicator / Report / DiaryEntry,缺 Asset / ChatTurn
├── DesignSystem/ ✅ Tokens + Components,沿用
└── Features/
├── Home/ ✅ HomeView 静态 UI,数据未接
├── Quick/A1-A3 🔧 待合并进 UnifiedCaptureFlow
├── Archive/B1-B4 🔧 B1 砍,B2 改用 VisionKit DocumentCamera
├── Record/RecordSheet ✅ 入口选择 UI
├── Trends/ ❌ 只有 placeholder
└── Me/ ❌ 只有 placeholder
待建:
├── AI/ ⚠️ AIRuntime + LLMSession + ModelStore + TokenChunk ✅;VLSession + Prompts/ ❌
├── Debug/DebugAIRunner.swift ✅ DEBUG-only AI 自检入口
├── Services/ ❌ CaptureService, AskService, TrendService, ReportCompareService
├── Persistence/FileVault.swift ✅ 原图加密目录管理
├── Security/AppLock.swift ❌ Face ID 启动锁
├── Features/Ask/ ❌ AskSheet (RAG 问答 UI)
├── Features/Archive/
│ ├── ArchiveListView ❌ C1 档案列表(分类 chip + 年份分组)
│ └── ReportDetailView ❌ C2 报告详情(三 Tab:原图/解读/指标 + 对比上次)
├── Features/Capture/
│ └── UnifiedCaptureFlow ❌ 替代 QuickCaptureFlow,状态机驱动 A1→VL→A2/B3
├── Features/Onboarding/ ❌ 首启动隐私承诺 + 模型下载
└── LiveActivity/ ❌ WidgetExtension target
```
---
## 9. 设计系统约束
- **不要新增颜色 token**。所有颜色走 `Tj.Palette.*` (sand / paper / ink / brick / leaf / line / text / text3)
- **不要新增字体大小**。走 `Font.tjTitle()` / `tjH2()` / `tjSerifBody()` / 系统 size
- **圆角走 `Tj.Radius.*`**,卡片走 `.tjCard()` modifier
- 按钮走 `TjPrimaryButton` / `TjGhostButton`
新加 View 时先看 `DesignSystem/Components.swift`,有现成的不要复刻。
---
## 10. 不能跨越的红线
写代码前必读:
1. **不引入云服务**——任何 SDK 都不行,包括崩溃上报、分析、灰度
2. **不自己实现密码学**——`.completeFileProtection` 已经够
3. **UI 不直接调 AIRuntime**——必须经过 Service
4. **AIRuntime 必须 actor 化**——禁止 class + lock
5. **VL/LLM prompt 必须有 few-shot + 失败回退**——不能让用户卡在 AI 错误屏
6. **新功能必须问"清单里有吗"**——清单外的功能(用药提醒、多 profile、暗黑模式、iCloud 同步……)默认不做,要做必须先讨论。**已加回的例外**:报告对比(16.1,§7.2)、症状追踪(Symptom @Model)、长期监测指标(MonitorMetric / IndicatorQuickSheet,W2)、个人资料(UserProfile,W2)
7. **不要在 6 周里重构现有 Tab/RecordSheet 骨架**——增量加东西,不要推倒重来
8. **报告详情(C2)与归档元信息编辑(B3)是两个 View**——B3 是 draft 编辑(写),C2 是 detail 浏览(读),不要合并复用主框架
---
## 11. 6 周时间表
| 周次 | 必交付 |
|---|---|
| W1 末 / W2 当前 | 项目结构、MLX 跑通 Qwen3-1.7B、首个 token 在设备吐出 |
| W2-W3 | AIRuntime + LLMSession,文字日记 + 基础 RAG 问答(打字机效果)(W2 进行中) |
| W3-W4 | VLSession + 统一拍照流程(单项 + 整份)、Asset / FileVault |
| W4 末 | **C1 ArchiveListView**(分类 chip + 年份分组,接 @Query) |
| W4-W5 | 趋势(Swift Charts + AI 解读)、**C2 ReportDetailView**(三 Tab + 重新解读) |
| W5 中 | **ReportCompareService** + C2 解读 Tab "对比上次" 区块 |
| W5 末 | Face ID、永久删除、首页时间线接入真数据、Live Activity(真机) |
| W6 | 模型管理页、首启动下载流程、UI polish、demo 视频、PPT |
**P0 新增项**(从 C 区视觉稿补回):C1 档案列表、C2 报告详情三 Tab、对比上次、关联到趋势、重新解读
**P1**(必须做完):Live Activity、分享文字摘要(9.4)、首启动隐私承诺页
**P2**(余力做):—— 任何 P2/P3 都暂时不做,清单里 11/12/13/14/15/17/18/19 全部 deferred(注意:16.1 报告对比已升 P0)
**砍 P1 决策顺序**(任何一周延期触发):Live Activity → Onboarding 简化 → 分享摘要 → 模型管理页 polish。**绝不动 C1/C2/对比上次**——视觉稿都做了,demo PPT 也要展示,这是核心卖点之一。
---
## 12. 评委 PPT 卖点排序(写代码时记住为什么这么做)
1. 影像档案系统(统一 VL 拍照 + 归档) — 核心创意
2. 100% 本地 + SME2 加速 — 技术亮点
3. 本地 RAG 长期记忆 — 端侧不可替代性
4. 隐私三件套(系统级加密 + Face ID + 永久删除) — 信任建立
5. AI 趋势解读 — 长期价值
6. Live Activity 实时 tok/s — 现场记忆点
每写一个功能,问自己:这条提升了上面哪一项?如果都没有,就别做。

112
scripts/build-launch.sh Executable file
View File

@@ -0,0 +1,112 @@
#!/usr/bin/env bash
set -euo pipefail
ROOT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
PROJECT="${PROJECT:-$ROOT_DIR/康康.xcodeproj}"
SCHEME="${SCHEME:-康康}"
APP_NAME="${APP_NAME:-$SCHEME}"
CONFIGURATION="${CONFIGURATION:-Debug}"
BUNDLE_ID="${BUNDLE_ID:-com.xuhuayong.kangkang}"
DERIVED_DATA_PATH="${DERIVED_DATA_PATH:-$ROOT_DIR/build/DerivedData}"
SIMULATOR_NAME="${SIMULATOR_NAME:-iPhone 16 Pro}"
SCREENSHOT_PATH="${SCREENSHOT_PATH:-$ROOT_DIR/build/screenshots/${SCHEME}-launch.png}"
require_tool() {
if ! command -v "$1" >/dev/null 2>&1; then
echo "error: required tool not found: $1" >&2
exit 1
fi
}
require_full_xcode() {
local developer_dir
developer_dir="$(xcode-select -p 2>/dev/null || true)"
if [[ "$developer_dir" != *"/Xcode.app/Contents/Developer"* ]]; then
cat >&2 <<EOF
error: active developer directory is not a full Xcode install:
${developer_dir:-<unset>}
Select Xcode before running this script:
sudo xcode-select -s /Applications/Xcode.app/Contents/Developer
EOF
exit 1
fi
}
extract_udid() {
sed -n 's/.*(\([0-9A-Fa-f-]\{36\}\)).*/\1/p' | head -n 1
}
find_simulator_udid() {
if [[ -n "${SIMULATOR_UDID:-}" ]]; then
echo "$SIMULATOR_UDID"
return
fi
local udid
udid="$(xcrun simctl list devices available | grep -F "$SIMULATOR_NAME" | extract_udid || true)"
if [[ -n "$udid" ]]; then
echo "$udid"
return
fi
udid="$(
xcrun simctl list devices available |
awk '/-- iOS / { in_ios = 1; next } /-- / { in_ios = 0 } in_ios && /iPhone/ { print; exit }' |
extract_udid || true
)"
if [[ -n "$udid" ]]; then
echo "$udid"
return
fi
echo "error: no available iOS simulator found. Install an iPhone simulator in Xcode." >&2
exit 1
}
main() {
require_tool xcode-select
require_tool xcodebuild
require_tool xcrun
require_full_xcode
local simulator_udid app_path
simulator_udid="$(find_simulator_udid)"
echo "Project: $PROJECT"
echo "Scheme: $SCHEME"
echo "Configuration: $CONFIGURATION"
echo "Simulator: ${SIMULATOR_UDID:-$SIMULATOR_NAME} ($simulator_udid)"
xcodebuild \
-project "$PROJECT" \
-scheme "$SCHEME" \
-configuration "$CONFIGURATION" \
-destination "id=$simulator_udid" \
-derivedDataPath "$DERIVED_DATA_PATH" \
build
app_path="$DERIVED_DATA_PATH/Build/Products/${CONFIGURATION}-iphonesimulator/${APP_NAME}.app"
if [[ ! -d "$app_path" ]]; then
echo "error: built app not found at $app_path" >&2
exit 1
fi
xcrun simctl boot "$simulator_udid" >/dev/null 2>&1 || true
xcrun simctl bootstatus "$simulator_udid" -b
if [[ "${OPEN_SIMULATOR:-1}" == "1" ]]; then
open -a Simulator --args -CurrentDeviceUDID "$simulator_udid"
fi
xcrun simctl install "$simulator_udid" "$app_path"
xcrun simctl launch "$simulator_udid" "$BUNDLE_ID"
mkdir -p "$(dirname "$SCREENSHOT_PATH")"
sleep "${SCREENSHOT_DELAY_SECONDS:-2}"
xcrun simctl io "$simulator_udid" screenshot "$SCREENSHOT_PATH"
echo "Launched $BUNDLE_ID"
echo "Screenshot: $SCREENSHOT_PATH"
}
main "$@"

View File

@@ -22,7 +22,7 @@ struct AboutView: View {
section(icon: "sparkles", title: String(appLoc: "这是什么")) { section(icon: "sparkles", title: String(appLoc: "这是什么")) {
paragraph( paragraph(
String(appLoc: "康康是一款以本地优先为设计原则的个人健康影像档案工具。") + String(appLoc: "康康是一款以本地优先为设计原则的个人健康随记工具。") +
String(appLoc: "你可以拍下体检报告、化验单和影像资料,图片与数据默认保存在本机;") + String(appLoc: "你可以拍下体检报告、化验单和影像资料,图片与数据默认保存在本机;") +
String(appLoc: "设备上的 AI 模型会尝试把专业指标转述为通俗说明,帮你记录并回顾自己的健康变化。") String(appLoc: "设备上的 AI 模型会尝试把专业指标转述为通俗说明,帮你记录并回顾自己的健康变化。")
) )
@@ -107,7 +107,7 @@ struct AboutView: View {
.font(.tjH2()) .font(.tjH2())
.foregroundStyle(Tj.Palette.text) .foregroundStyle(Tj.Palette.text)
Text("本地优先的个人健康影像档案") Text("本地优先的个人健康随记")
.font(.system(size: 13)) .font(.system(size: 13))
.foregroundStyle(Tj.Palette.text2) .foregroundStyle(Tj.Palette.text2)

View File

@@ -5609,24 +5609,24 @@
} }
} }
}, },
"康康是一款以本地优先为设计原则的个人健康影像档案工具。" : { "康康是一款以本地优先为设计原则的个人健康随记工具。" : {
"localizations" : { "localizations" : {
"en" : { "en" : {
"stringUnit" : { "stringUnit" : {
"state" : "translated", "state" : "translated",
"value" : "Kangkang is a personal health imaging archive tool designed with a local-first principle." "value" : "Kangkang is a personal health journal tool designed with a local-first principle."
} }
}, },
"ja" : { "ja" : {
"stringUnit" : { "stringUnit" : {
"state" : "translated", "state" : "translated",
"value" : "Kangkangは、ローカルファーストを設計原則とする個人向け健康画像アーカイブツールです。" "value" : "Kangkangは、ローカルファーストを設計原則とする個人向け健康ジャーナルツールです。"
} }
}, },
"ko" : { "ko" : {
"stringUnit" : { "stringUnit" : {
"state" : "translated", "state" : "translated",
"value" : "Kangkang은 로컬 우선을 설계 원칙으로 하는 개인 건강 영상 아카이브 도구입니다." "value" : "Kangkang은 로컬 우선을 설계 원칙으로 하는 개인 건강 저널 도구입니다."
} }
} }
} }
@@ -8138,24 +8138,24 @@
} }
} }
}, },
"本地优先的个人健康影像档案" : { "本地优先的个人健康随记" : {
"localizations" : { "localizations" : {
"en" : { "en" : {
"stringUnit" : { "stringUnit" : {
"state" : "translated", "state" : "translated",
"value" : "Local-first personal health imaging archive" "value" : "Local-first personal health journal"
} }
}, },
"ja" : { "ja" : {
"stringUnit" : { "stringUnit" : {
"state" : "translated", "state" : "translated",
"value" : "ローカル優先の個人健康画像アーカイブ" "value" : "ローカル優先の個人健康ジャーナル"
} }
}, },
"ko" : { "ko" : {
"stringUnit" : { "stringUnit" : {
"state" : "translated", "state" : "translated",
"value" : "로컬 우선 개인 건강 영상 아카이브" "value" : "로컬 우선 개인 건강 저널"
} }
} }
} }
@@ -12207,4 +12207,4 @@
} }
}, },
"version" : "1.0" "version" : "1.0"
} }

View File

@@ -179,7 +179,7 @@ actor CaptureService {
let summary = (dict["summary"] as? String) ?? "" let summary = (dict["summary"] as? String) ?? ""
let pages = (dict["page_count"] as? Int) ?? pageCount let pages = (dict["page_count"] as? Int) ?? pageCount
let indicatorsRaw = (dict["indicators"] as? [[String: Any]]) ?? [] let indicatorsRaw = arrayValue(dict, keys: ["indicators", "indicator", "items", "指标", "指标列表", "项目"])
let indicators: [ParsedReport.ParsedIndicator] = indicatorsRaw.compactMap { let indicators: [ParsedReport.ParsedIndicator] = indicatorsRaw.compactMap {
parseIndicator($0) parseIndicator($0)
} }
@@ -212,7 +212,7 @@ actor CaptureService {
// :{"indicators":[...]} [...]( key) // :{"indicators":[...]} [...]( key)
let indicatorsRaw: [[String: Any]] let indicatorsRaw: [[String: Any]]
if let dict = obj as? [String: Any] { if let dict = obj as? [String: Any] {
indicatorsRaw = (dict["indicators"] as? [[String: Any]]) ?? [] indicatorsRaw = arrayValue(dict, keys: ["indicators", "indicator", "items", "指标", "指标列表", "项目"])
} else if let arr = obj as? [[String: Any]] { } else if let arr = obj as? [[String: Any]] {
indicatorsRaw = arr indicatorsRaw = arr
} else { } else {
@@ -335,18 +335,100 @@ actor CaptureService {
} }
private static func parseIndicator(_ d: [String: Any]) -> ParsedReport.ParsedIndicator? { private static func parseIndicator(_ d: [String: Any]) -> ParsedReport.ParsedIndicator? {
guard let name = (d["name"] as? String)?.trimmingCharacters(in: .whitespaces), guard let name = stringValue(d, keys: ["name", "item", "indicator", "test", "项目", "指标", "指标名", "指标名称", "检查项目", "检验项目"])?.trimmingCharacters(in: .whitespaces),
!name.isEmpty else { return nil } !name.isEmpty else { return nil }
let value: String let value: String
if let v = d["value"] as? String { value = v } if let v = stringValue(d, keys: ["value", "result", "reading", "结果", "数值", "检测值", "测定值"]) { value = v }
else if let v = d["value"] as? NSNumber { value = v.stringValue }
else { value = "" } else { value = "" }
let unit = (d["unit"] as? String) ?? "" let unit = stringValue(d, keys: ["unit", "单位"]) ?? ""
let range = (d["range"] as? String) ?? "" let range = stringValue(d, keys: ["range", "reference", "reference_range", "ref", "参考", "参考值", "参考范围", "正常范围"]) ?? ""
let statusRaw = (d["status"] as? String)?.lowercased() ?? "normal" let statusRaw = stringValue(d, keys: ["status", "flag", "abnormal", "异常", "提示", "标记"])
let status = IndicatorStatus(rawValue: statusRaw) ?? .normal let status = parseIndicatorStatus(raw: statusRaw, value: value, range: range)
return .init(name: name, value: value, unit: unit, range: range, status: status) return .init(name: name, value: value, unit: unit, range: range, status: status)
} }
private static func stringValue(_ d: [String: Any], keys: [String]) -> String? {
for key in keys {
if let s = d[key] as? String {
return s
}
if let n = d[key] as? NSNumber {
return n.stringValue
}
}
return nil
}
private static func arrayValue(_ d: [String: Any], keys: [String]) -> [[String: Any]] {
for key in keys {
if let arr = d[key] as? [[String: Any]] {
return arr
}
}
return []
}
private static func parseIndicatorStatus(raw: String?, value: String, range: String) -> IndicatorStatus {
let normalized = raw?
.trimmingCharacters(in: .whitespacesAndNewlines)
.lowercased() ?? ""
if ["high", "h", "hi", "above", "up", "", "", "+", "偏高", "", "增高", "升高", "偏高↑", "h↑"].contains(normalized) {
return .high
}
if ["low", "l", "lo", "below", "down", "", "", "-", "偏低", "", "降低", "偏低↓", "l↓"].contains(normalized) {
return .low
}
if ["normal", "n", "ok", "正常", "阴性", "无异常"].contains(normalized) {
return .normal
}
return inferStatus(value: value, range: range) ?? .normal
}
private static func inferStatus(value: String, range: String) -> IndicatorStatus? {
guard let v = firstNumber(in: value) else { return nil }
let compact = range
.replacingOccurrences(of: "", with: "-")
.replacingOccurrences(of: "", with: "-")
.replacingOccurrences(of: "", with: "-")
.replacingOccurrences(of: "~", with: "-")
.replacingOccurrences(of: "", with: "-")
.trimmingCharacters(in: .whitespacesAndNewlines)
guard !compact.isEmpty else { return nil }
let numbers = numbers(in: compact)
if compact.contains("<") || compact.contains("") || compact.contains("") {
guard let upper = numbers.first else { return nil }
return v > upper ? .high : .normal
}
if compact.contains(">") || compact.contains("") || compact.contains("") {
guard let lower = numbers.first else { return nil }
return v < lower ? .low : .normal
}
if numbers.count >= 2 {
let lower = numbers[0]
let upper = numbers[1]
if v < lower { return .low }
if v > upper { return .high }
return .normal
}
return nil
}
private static func firstNumber(in text: String) -> Double? {
numbers(in: text).first
}
private static func numbers(in text: String) -> [Double] {
let pattern = #"-?\d+(?:\.\d+)?"#
guard let regex = try? NSRegularExpression(pattern: pattern) else { return [] }
let ns = text as NSString
let range = NSRange(location: 0, length: ns.length)
return regex.matches(in: text, range: range).compactMap {
Double(ns.substring(with: $0.range))
}
}
} }
// MARK: - Report CaptureService (MainActor ) // MARK: - Report CaptureService (MainActor )

View File

@@ -100,6 +100,42 @@ struct CaptureServiceJSONTests {
#expect(parsed.indicators.first?.status == .normal) #expect(parsed.indicators.first?.status == .normal)
} }
@Test func parsesRegionJSONWithChineseKeysAndArrowStatus() throws {
let raw = """
{"指标":[{"项目":"尿酸","结果":"486","单位":"μmol/L","参考范围":"208 - 428","异常":""}]}
"""
let indicators = try CaptureService.parseIndicatorsJSON(raw)
#expect(indicators.count == 1)
#expect(indicators.first?.name == "尿")
#expect(indicators.first?.value == "486")
#expect(indicators.first?.unit == "μmol/L")
#expect(indicators.first?.range == "208 - 428")
#expect(indicators.first?.status == .high)
}
@Test func parsesReportJSONWithChineseIndicatorArrayKey() throws {
let raw = """
{"title":"t","type":"lab","report_date":"2026-05-01","指标":[{"项目":"尿酸","结果":"486","单位":"μmol/L","参考范围":"208 - 428","异常":"偏高"}]}
"""
let parsed = try CaptureService.parseReportJSON(raw)
#expect(parsed.indicators.count == 1)
#expect(parsed.indicators.first?.name == "尿")
#expect(parsed.indicators.first?.status == .high)
}
@Test func infersStatusFromValueAndReferenceRangeWhenStatusMissing() throws {
let raw = """
{"indicators":[
{"name":"低密度脂蛋白","value":"3.84","unit":"mmol/L","range":"< 3.40"},
{"name":"白细胞","value":"2.8","unit":"10^9/L","range":"3.5 - 9.5"}
]}
"""
let indicators = try CaptureService.parseIndicatorsJSON(raw)
#expect(indicators.count == 2)
#expect(indicators[0].status == .high)
#expect(indicators[1].status == .low)
}
@Test func badReportDateFallsBackToNow() throws { @Test func badReportDateFallsBackToNow() throws {
let raw = """ let raw = """
{"title":"t","type":"lab","report_date":"昨天","institution":"","page_count":1,"summary":"","indicators":[]} {"title":"t","type":"lab","report_date":"昨天","institution":"","page_count":1,"summary":"","indicators":[]}