@harmonyos-arkts/d2h 0.1.19 → 0.1.21

package/skills/d2h-spec-plan/references/work-packets.md

@@ -0,0 +1,61 @@
+# 自足工作包、共享状态与实际接口
+
+创建或细化当前轮次、处理开工缺口、设计跨 Spec 状态流时读取。计划字段归 pages.json；接口回执归 execution.json。context 裁剪输出，不另建可编辑工单真源。
+
+## 开工就绪
+
+新计划使用 `work_packet_policy: "required-v1"`。每个 Spec 提供：
+
+```json
+{
+  "implementation_brief": {
+    "behavior_notes": ["输入、状态转移、结果与必要失败路径"],
+    "invariants": ["本次行为不能破坏的约束"],
+    "known_pitfalls": ["从源码或实际失败确认的风险"],
+    "open_questions": []
+  },
+  "testing": {
+    "layers": [{"check_id": "login/basic", "layer": "device", "reason": "真实鉴权与导航"}]
+  }
+}
+```
+
+behavior_notes 非空；其余三个数组必须存在，无适用内容经核对后写 []。open_questions 只放阻止可靠实现的未决问题，非阻塞背景放 behavior_notes。testing.layers 唯一且完整覆盖本 Spec 的 checks；layer 为 static/logic/integration/device，reason 非空。可暂存不完整计划，但 derive.readiness_gaps 阻止相关成员进入窗口，不能夹带进同页批次。
+
+工作包复用 scope、evidence_refs、owned_paths、requires/provides 和 checks。AI 按任务补齐 UI 入口与路由参数、数据表/键/副作用、逻辑模块及目标文件；纯后台/算法能力不虚构页面。源码仍是行为细节依据，摘要用于定位与交接。远期延期需求不提前写大工单。
+
+## 跨 Spec 共享状态
+
+仅在多个 Spec 实际共享状态时写顶层 shared_contracts，不为每个功能强建基础设施。
+
+```json
+{
+  "state_id": "session",
+  "capability": "session.basic",
+  "producer_spec_id": "login-basic",
+  "consumer_spec_ids": ["home-basic"],
+  "channel": "SessionStore 单一状态通道",
+  "payload": "令牌、用户标识及匿名/有效/过期状态",
+  "initialization": "登录成功持久化；首页订阅并读取快照",
+  "lifecycle": "注销或过期清空并通知消费者",
+  "checks": ["login/persist-session", "home/observe-session"]
+}
+```
+
+producer 提供指定 capability，每个 consumer 必须 requires 它；checks 包含每个参与 Spec 的实际检查，不得越界。channel/payload/initialization/lifecycle 缺失时参与者不能开工。登录和首页 UI 都存在不能证明状态已传递；必须明确谁写、谁读、何时生效及失效，并用消费者断言覆盖状态变化。
+
+## 真实接口交接
+
+提供者实现并形成当前候选后，主 Agent 阅读代码核对实际导出符号、参数/返回及错误语义，以 interface 事件记录代码文件及哈希、symbols、summary、reviewer。运行时确认由调用方按公共 `d2h workflow contract execution` 执行；本 Skill 不记录事件。这不是预先编造的签名，也不是运行证据。提供者只有接口回执有效、当前构建和即时契约测试有效，才解锁消费者。
+
+candidate 或接口文件变化、failed/invalidate/repair 会撤销接口有效性。context 不交接过期接口；共享契约或工作包语义通过 reconcile 修订并重置相关成员，保留历史。程序检查文件身份，不解析符号真实性或证明实际调用关系，主 Agent 负责语义核验。
+
+## 分层验证与定向修复
+
+分层是选择验证成本的计划，不改变通过标准。当前正式 test evidence 仍只支持受管 Python 设备执行；静态检查、算法向量和服务探针可前置发现问题，但尚无 CLI 证据适配，不能用于正式 PASS。鉴权、启动、共享契约和其他高风险行为仍即时验证，低风险成员按窗口合批。
+
+只修复并重跑失败的准备步骤，避免重新提取已完成需求；同一候选可共享构建安装和真实结果。主 Agent 对高风险且缺证据的行为优先检查反例，如失效令牌、重复提交、返回后状态丢失，不为每项重复启动多轮主观评审。候选变化仍保守撤销旧验证，影响范围复用需后续 CLI 协同。
+
+## 与完整 Spec 的关系
+
+implementation_brief 是结构化计划/派发摘要，不替代详细行为、操作、证据与逐 check 断言。规划先提供已核验的概要及缺口；生成方按公共 `d2h workflow contract generation` 展开正文，缺少实施字段时提出修订建议。新策略额外核对逐成员生成绑定，但 readiness_gaps 仍不验证 Markdown 语义；AI/人工须审查。worker 根据固定 spec_id 补读详细文档，不根据摘要猜测。

package/skills/d2h-utg-ai-generate/references/click-paths-schema.json

@@ -1,7 +1,7 @@
 {
   "$schema": "https://json-schema.org/draft/2020-12/schema",
-  "$id": "https://hometrans.local/schemas/click-paths-v2.json",
-  "title": "HomeTrans deterministic click paths",
+  "$id": "https://d2h.local/schemas/click-paths-v2.json",
+  "title": "d2h deterministic click paths",
   "type": "object",
   "required": ["root", "paths", "unreachable"],
   "properties": {

package/skills/d2h-utg-ai-generate/references/page-schema.json

@@ -1,7 +1,7 @@
 {
   "$schema": "https://json-schema.org/draft/2020-12/schema",
-  "$id": "https://hometrans.local/schemas/page-plan-v1.json",
-  "title": "HomeTrans UTG-directed runtime capture plan",
+  "$id": "https://d2h.local/schemas/page-plan-v1.json",
+  "title": "d2h UTG-directed runtime capture plan",
   "type": "object",
   "required": ["schema_version", "source", "summary", "pages"],
   "properties": {

package/skills/d2h-utg-ai-generate/references/utg-schema.json

@@ -1,7 +1,7 @@
 {
   "$schema": "https://json-schema.org/draft/2020-12/schema",
-  "$id": "https://hometrans.local/schemas/utg-v2.json",
-  "title": "HomeTrans source-derived UTG",
+  "$id": "https://d2h.local/schemas/utg-v2.json",
+  "title": "d2h source-derived UTG",
   "type": "object",
   "required": ["directed", "multigraph", "graph", "nodes", "links"],
   "properties": {

package/skills/d2h-utg-static-generate/SKILL.md

@@ -19,14 +19,14 @@ description: 迁移主流程的 Android 静态分析入口，在 Skill 内调用
    droid2hmos analyze android <android-source>
    ```
 
-2. 确认本次生成的 `<migration-workspace>/spec/index.json`、`spec/facts/` 和 `appgraph/app-graph.json` 存在，核对源码与分析结果。复用这份 AppGraph，不再执行第二次 `appgraph build`：
+2. 确认本次生成的 `<migration-workspace>/spec/index.json`、`spec/facts/`（含 `facts/native-libs.json` 原生库分片）和 `appgraph/app-graph.json` 存在，核对源码与分析结果。复用这份 AppGraph，不再执行第二次 `appgraph build`：
 
    ```bash
    python <skill>/scripts/generate.py --project-root <android-source> --workspace <migration-workspace>/utg --appgraph <migration-workspace>/appgraph/app-graph.json
    ```
 
 3. 按下文完成源码核验、独立 AI 修补和统一重建。已有 UTG 工作区需显式使用 `--force`；源码或 AppGraph 变化时重新核验旧修补，不能沿用过期指纹。
-4. 将 Facts 位置、受管分析结果、UTG 产物及未解析项交回主 Agent，由主 Agent 接续报告初始化与 Task Plan。UTG discoveries 不替代 Kernel 的 correction/review，也不替代受管分析证据。
+4. 将 Facts 位置、受管分析结果、UTG 产物及未解析项交回主 Agent，由主 Agent 接续报告初始化与 Task Plan。`facts/native-libs.json` 的 `present`/`usageCase`/`libraries`/`loadCalls` 原样转交（其处置归 `d2h-native-library-build`，本 Skill 不做 ELF/符号深检）。UTG discoveries 不替代 Kernel 的 correction/review，也不替代受管分析证据。
 
 Android 分析不运行 Gradle、不启用 `--resolve-dependencies`，不需要 APK 或 Android 设备。已有图缺失时报告分析产物不完整，不在主流程模式中静默切换到独立建图。
 

package/skills/managing-d2h-toolchain/environment-requirements.json

@@ -8,7 +8,7 @@
     "java": null,
     "python-json5": null,
     "appgraph": { "minimum": "0.0.11", "maximumExclusive": "0.0.12" },
-    "droid2hmos": { "minimum": "0.0.12", "maximumExclusive": "0.0.13" },
+    "droid2hmos": { "minimum": "0.0.17", "maximumExclusive": "0.0.18" },
     "hvigor": null,
     "ohpm": null,
     "devecocli": null

package/skills/managing-d2h-toolchain/toolchain.json

@@ -9,7 +9,7 @@
     },
     "droid2hmos": {
       "package": "@harmonyos-arkts/droid2hmos",
-      "version": "0.0.12",
+      "version": "0.0.17",
       "capabilities": { "analyze-android": ["--help"], "preflight": ["--help"], "status": ["--help"] }
     }
   }

package/skills/preparing-migration-workspace/SKILL.md

@@ -9,10 +9,38 @@ description: 在 Android 转 HarmonyOS 迁移开始或恢复时，通过 droid2h
 
 工具安装、精确版本和升级由 `managing-d2h-toolchain` 负责。本 Skill 只消费已检查的 CLI；发现版本或能力不符时返回该 Skill 处理，不运行 npm。
 
-只调用 `droid2hmos` 的 `status`、`preflight`、`config init/validate` 和按需的 `device check`。不要直接调用 Node 脚本、Gradle、Hvigor、Hypium、ADB、HDC 或 AppGraph 分析命令，也不要在 `ohos/` 尚不存在时猜测 HarmonyOS 构建配置。
+只调用 `droid2hmos` 的 `status`、`preflight`、`config init/validate`、`progress init/update` 和按需的 `device check`。不要直接调用 Node/Python 脚本、Gradle、Hvigor、Hypium、ADB、HDC 或 AppGraph 分析命令，也不要在 `ohos/` 尚不存在时猜测 HarmonyOS 构建配置。
 
 迁移产物（`.migration/`）统一位于 Android 工程父目录下的 `<工程名>-migration/.migration/`，由 CLI 创建和维护，不写入 Android 工程内部。下文所有 `.migration/...` 路径均相对该迁移工作区根目录。
 
+## 阶段进度文件
+
+CLI 确认迁移工作区且 `.migration/` 已存在后，立即初始化阶段进度文件：
+
+```text
+droid2hmos progress init <ANDROID_PROJECT>
+```
+
+该命令原子创建 `.migration/progress.json`；恢复迁移时若文件已存在，只校验结构并保留全部已有时间和状态，禁止覆盖重置。文件固定包含以下七个有序阶段：
+
+1. 初始化工程（`workspace-initialization`）
+2. 功能清单生成（`feature-inventory`）
+3. Android 工程分析（`android-analysis`）
+4. HarmonyOS 工程架构设计（`harmony-architecture`）
+5. spec 设计与生成（`spec-design-generation`）
+6. 代码转换（`code-conversion`）
+7. 测试与交付（`testing-delivery`）
+
+每个阶段记录 `status`、`startedAt`、`endedAt`、`completed`、`lastUpdatedAt`、`details` 和 `blockers`；顶层 `summary` 汇总总数及 `pending/in_progress/blocked/completed` 数量和当前阶段。所有时间使用带时区偏移的 ISO 8601。初始化时“初始化工程”为 `in_progress`，其余阶段为 `pending`。
+
+本 Skill 只更新“初始化工程”阶段。Preflight 和配置准备达到 `baselineReady=true` 时执行：
+
+```text
+droid2hmos progress update <ANDROID_PROJECT> --stage workspace-initialization --status completed --details "workspace, configuration, and non-device baseline are ready"
+```
+
+若工程定位、配置或必要工具链使 `allowedNonDeviceWork=false`，将该阶段更新为 `blocked` 并用 `--blocker` 原样记录 CLI 返回的阻塞原因；设备单独不可用不阻止初始化阶段完成。后续编排者在对应阶段开始、完成或受阻时用同一 `update` 命令更新 `progress.json`，不得直接手工改写、删除阶段或伪造时间。
+
 ## 输入与事实来源
 
 输入必须包含 Android Gradle 工程根目录。根目录应有 Gradle 标记，应用模块中应有 `AndroidManifest.xml`。
@@ -112,7 +140,7 @@ droid2hmos config validate <ANDROID_PROJECT>
   ],
   "allowedNonDeviceWork": true,
   "allowedActions": [],
-  "evidence": [".migration/preflight-report.json"],
+  "evidence": [".migration/preflight-report.json", ".migration/progress.json"],
   "suggestedRemediation": []
 }
 ```
@@ -128,4 +156,6 @@ droid2hmos config validate <ANDROID_PROJECT>
 - `evidence`：本次读取或生成的状态、配置和 Preflight 报告；
 - `suggestedRemediation`：CLI 给出的可执行补救，不臆造成功结论。
 
+结构化结果的 `evidence` 必须包含本次读取或初始化的 `.migration/progress.json`。
+
 本 Skill 不推进外部状态机、不更新迁移报告、不创建 checkpoint、不调用 `complete`，也不决定定向测试或全量回归。调用方根据上述结果决定下一步。

package/workflow/artifact_bindings.py

@@ -0,0 +1,224 @@
+"""Read-only bindings between analysis reports, versioned requirements and dashboard metadata."""
+import hashlib
+import json
+from datetime import datetime, timezone
+from pathlib import Path
+
+
+FEATURE_LIST_REQUIRED = ["trigger", "result", "priority", "evidence"]
+
+
+def validate_legacy_report(data):
+    errors = []
+    for field in ("project", "feature_list", "summary"):
+        if field not in data:
+            errors.append("缺少必填字段: {}".format(field))
+    if errors:
+        return errors
+    if not isinstance(data["project"], str) or not data["project"].strip():
+        errors.append("project 必须是非空字符串")
+    if not isinstance(data["feature_list"], list) or not data["feature_list"]:
+        errors.append("feature_list 必须是非空数组")
+    else:
+        for idx, row in enumerate(data["feature_list"], 1):
+            if not isinstance(row, dict):
+                errors.append("feature_list 第 {} 行不是对象".format(idx))
+                continue
+            for field in FEATURE_LIST_REQUIRED:
+                value = str(row.get(field, "") or "").strip()
+                if not value:
+                    errors.append("feature_list 第 {} 行缺少必填字段 {}".format(idx, field))
+            if str(row.get("priority", "")).strip() not in ("P0", "P1", "P2"):
+                errors.append("feature_list 第 {} 行 priority 必须是 P0/P1/P2".format(idx))
+    if not isinstance(data["summary"], list) or not data["summary"]:
+        errors.append("summary 必须是非空数组")
+    return errors
+
+
+def validate_inventory_contract(data):
+    """Validate the exchanged artifact, not a producer's source-analysis workflow."""
+    errors = []
+    def require(value, message):
+        if not value:
+            errors.append(message)
+    require(data.get('schema_version') == '1.1', 'schema_version must be 1.1')
+    require(data.get('max_depth') == 4, 'max_depth must be 4')
+    repositories = data.get('repositories')
+    require(isinstance(repositories, list) and bool(repositories), 'repositories must be non-empty')
+    ids = []
+    for repo in repositories if isinstance(repositories, list) else []:
+        if not isinstance(repo, dict):
+            errors.append('repository must be an object')
+            continue
+        identity = repo.get('id')
+        require(isinstance(identity, str) and bool(identity), 'repository id required')
+        if isinstance(identity, str):
+            ids.append(identity)
+        require(isinstance(repo.get('path'), str) and Path(repo['path']).is_absolute(), 'repository path must be absolute')
+    require(len(set(ids)) == len(ids), 'duplicate repository id')
+    rows = data.get('features')
+    require(isinstance(rows, list) and bool(rows), 'features must be non-empty')
+    seen = set()
+    for i, row in enumerate(rows if isinstance(rows, list) else []):
+        label = f'features[{i}]'
+        if not isinstance(row, dict):
+            errors.append(label + ' must be an object')
+            continue
+        path = row.get('path')
+        valid_path = isinstance(path, list) and 1 <= len(path) <= 4 and all(isinstance(p, str) and bool(p.strip()) for p in path)
+        require(valid_path, label + '.path must contain 1-4 non-empty names')
+        context = row.get('entry_context')
+        require(isinstance(context, str) and bool(context.strip()), label + '.entry_context required')
+        if valid_path and isinstance(context, str):
+            key = (context, tuple(path))
+            require(key not in seen, label + ': duplicate path/context')
+            seen.add(key)
+        require(row.get('release_stage') in ('mvp', 'core', 'full', 'uncertain'), label + '.release_stage invalid')
+        require(isinstance(row.get('release_rationale'), str) and bool(row['release_rationale'].strip()), label + '.release_rationale required')
+        implementations = row.get('implementations')
+        require(isinstance(implementations, dict) and set(implementations) == set(ids), label + '.implementations must match repositories')
+        for repo_id, impl in implementations.items() if isinstance(implementations, dict) else []:
+            if not isinstance(impl, dict):
+                errors.append(label + ': implementation must be an object')
+                continue
+            require(impl.get('status') in ('implemented', 'partial', 'mock', 'unreachable', 'absent', 'uncertain'), label + ': status invalid')
+            require(impl.get('confidence') in ('high', 'medium', 'low'), label + ': confidence invalid')
+            technology = impl.get('technology_stack')
+            require(isinstance(technology, list) and all(isinstance(t, str) for t in technology), label + ': technology_stack must be strings')
+            require(isinstance(impl.get('evidence', []), list), label + ': evidence must be an array')
+            for field in ('external_dependencies', 'android_permissions'):
+                section = impl.get(field)
+                require(isinstance(section, dict), label + ': ' + field + ' must be an object')
+                if not isinstance(section, dict):
+                    continue
+                require(section.get('status') in ('present', 'none', 'uncertain', 'not-applicable'), label + ': technical status invalid')
+                items = section.get('items', [])
+                require(isinstance(items, list) and all(isinstance(item, dict) and isinstance(item.get('name'), str) and bool(item['name']) for item in items), label + ': technical items invalid')
+                if field == 'external_dependencies':
+                    require(isinstance(items, list) and all(isinstance(item, dict) and item.get('kind') in ('sdk', 'so') for item in items), label + ': dependency kind invalid')
+    return errors
+
+
+def inventory_view(data):
+    """Read-only presentation projection; source references keep /features/N."""
+    result = {'project': data.get('title', ''), 'feature_list': [], 'summary': [],
+              'tech_stack': [], 'sdk_dependencies': [], 'permissions': [], 'user_inputs': [],
+              'uncovered': [{'component': 'inventory question', 'reason': str(q), 'evidence': ''}
+                            for q in data.get('open_questions', [])]}
+    for i, feature in enumerate(data['features']):
+        evidence = []
+        implementations = feature.get('implementations', {})
+        for repo_id, impl in implementations.items():
+            evidence.extend(f"{repo_id}:{e.get('path', '')}:{e.get('line_start', '')}" for e in impl.get('evidence', []))
+            for tech in impl.get('technology_stack', []):
+                result['tech_stack'].append({'item': tech, 'conclusion': repo_id, 'evidence': '; '.join(evidence), 'impact': ' / '.join(feature['path'])})
+            for dep in impl.get('external_dependencies', {}).get('items', []):
+                result['sdk_dependencies'].append({'sdk': dep['name'], 'category': dep['kind'],
+                    'harmony_support': '待确认', 'alternative': dep.get('purpose', ''),
+                    'evidence': f"/features/{i}/implementations/{repo_id}/external_dependencies", 'provenance': dep.get('provenance', 'unknown')})
+        result['feature_list'].append({'trigger': feature.get('trigger', ''), 'result': feature.get('effect', ''),
+            'condition': ' / '.join(feature.get('conditions', [])), 'carrier': ' / '.join(feature['path']),
+            'priority': feature.get('release_stage', 'uncertain'), 'note': feature.get('release_rationale', ''),
+            'evidence': '; '.join(evidence), 'implementations': implementations})
+    return result
+
+
+def load_report(path):
+    path = Path(path).resolve()
+    raw = path.read_bytes()
+    data = json.loads(raw)
+    if not isinstance(data, dict):
+        raise ValueError('feature-analysis report must be an object')
+    if data.get('schema_version') == '1.1':
+        errors = validate_inventory_contract(data)
+        if errors:
+            raise ValueError('inventory validation failed: ' + '; '.join(errors))
+        if not data.get('features'):
+            raise ValueError('inventory requires at least one feature for migration planning')
+        return {'path': str(path), 'sha256': hashlib.sha256(raw).hexdigest(), 'pointer_prefix': '/features/',
+                'source_schema': 'inventory/1.1', 'inventory': data, 'data': inventory_view(data)}
+    errors = validate_legacy_report(data)
+    if errors:
+        raise ValueError('; '.join(errors))
+    fingerprint = data.get('fingerprint')
+    if not isinstance(fingerprint, dict) or not isinstance(fingerprint.get('value'), str) or not fingerprint['value'].strip() or fingerprint.get('algorithm') == 'none':
+        raise ValueError('feature-analysis source fingerprint required; rerun analysis')
+    for key in ('tech_stack', 'sdk_dependencies', 'user_inputs', 'permissions', 'uncovered'):
+        if not isinstance(data.get(key, []), list) or any(not isinstance(row, dict) for row in data.get(key, [])):
+            raise ValueError('feature-analysis ' + key + ' must be an object array')
+    return {'path': str(path), 'sha256': hashlib.sha256(raw).hexdigest(), 'pointer_prefix': '/feature_list/', 'data': data}
+
+
+
+def validate_links(features):
+    errors = []
+    for source in features.get('sources', []):
+        if not isinstance(source, dict) or source.get('kind') not in ('inventory', 'source') or not isinstance(source.get('locator'), str) or Path(source['locator']).name not in ('feature-analysis.json', 'inventory.json'):
+            continue
+        try:
+            if not Path(source['locator']).is_absolute():
+                raise ValueError('feature-analysis locator must be absolute')
+            report = load_report(source['locator'])
+            if source.get('revision') != 'sha256:' + report['sha256']:
+                raise ValueError('feature-analysis report hash changed; revise upstream plan before Spec dispatch')
+            expected = {report['pointer_prefix'] + str(i) for i in range(len(report['data']['feature_list']))}
+            seen = set()
+            for feature in features.get('features', []):
+                for ref in feature.get('source_refs', []):
+                    if ref.get('source_id') == source['source_id']:
+                        pointer = ref.get('ref')
+                        if pointer not in expected:
+                            raise ValueError('feature-analysis ref must identify an existing ' + report['pointer_prefix'] + 'N row')
+                        seen.add(pointer)
+            if seen != expected:
+                raise ValueError('every analysis feature row needs explicit planning coverage, including deferred/excluded/unresolved')
+        except (OSError, ValueError, KeyError, TypeError) as error:
+            errors.append(str(error))
+    return errors
+
+
+def dashboard_report(path, features):
+    report = load_report(path)
+    sources = [s for s in features['sources'] if s.get('kind') in ('inventory', 'source') and Path(s.get('locator', '')).resolve() == Path(report['path']) and s.get('revision') == 'sha256:' + report['sha256']]
+    if not sources:
+        raise ValueError('--feature-analysis must be bound by the current features.json source path and report hash')
+    errors = validate_links(features)
+    if errors:
+        raise ValueError('; '.join(errors))
+    data = report['data']
+    overview = {'schema': 'd2h-overview/v1', 'generated_at': datetime.now(timezone.utc).isoformat(),
+                'source_revision': 'analysis-report-sha256:' + report['sha256'],
+                'sources': [{'source_id': 'feature-analysis', 'path': report['path'], 'sha256': report['sha256']}],
+                'tech_stack': [], 'dependencies': [], 'tools': [], 'blockers': []}
+    refs = ['feature-analysis']
+    stack_path = Path(report['path']).with_name('stack.json')
+    if report.get('source_schema') != 'inventory/1.1' and stack_path.exists():
+        stack_raw = stack_path.read_bytes()
+        stack = json.loads(stack_raw)
+        fp = data['fingerprint']
+        if not isinstance(stack, dict) or any(stack.get('fingerprint', {}).get(k) != fp.get(k) for k in ('algorithm', 'value')):
+            raise ValueError('stack.json fingerprint differs from feature-analysis; rerun consistent analysis')
+        languages = stack.get('languages')
+        if not isinstance(languages, dict) or any(not isinstance(row, dict) or any(type(row.get(k)) is not int or row[k] < 0 for k in ('files', 'lines')) for row in languages.values()):
+            raise ValueError('stack.json language counts invalid')
+        overview['sources'].append({'source_id': 'analysis-stack', 'path': str(stack_path), 'sha256': hashlib.sha256(stack_raw).hexdigest()})
+        overview['code'] = {'files': sum(row['files'] for row in languages.values()), 'lines': sum(row['lines'] for row in languages.values()),
+            'languages': [dict(name=name, files=row['files'], lines=row['lines']) for name, row in languages.items()],
+            'method': 'legacy stack report source extensions and physical lines; completeness not independently verified',
+            'source_refs': ['analysis-stack']}
+
+    for row in data.get('tech_stack', []):
+        overview['tech_stack'].append({'name': row.get('item') or '分析项未命名', 'scope': str(row.get('conclusion', '')), 'evidence': row.get('evidence', ''), 'impact': row.get('impact', ''), 'source_refs': refs})
+    for i, row in enumerate(data.get('sdk_dependencies', [])):
+        name = str(row.get('sdk') or '未命名依赖')
+        category = str(row.get('category', '')).lower()
+        kind = 'so' if 'so' in category or name.lower().endswith('.so') else 'sdk' if 'sdk' in category or 'sdk' in name.lower() else 'library'
+        overview['dependencies'].append({'name': name, 'kind': kind, 'platform': 'Android → HarmonyOS', 'provision': 'unknown',
+            'status': '报告适配结论：' + str(row.get('harmony_support') or '待确认'),
+            'reason': '尚未确认自编译或供应商产物处置；' + str(row.get('alternative', '')),
+            'evidence': row.get('evidence', ''), 'analysis_ref': '/sdk_dependencies/' + str(i), 'source_refs': refs})
+    for row in data.get('user_inputs', []):
+        overview['blockers'].append({'name': row.get('item') or '用户待提供项', 'category': row.get('kind') or '接入准备',
+            'status': 'open', 'reason': row.get('reason') or '报告记录待提供，实际准备状态待复核',
+            'action': row.get('usage') or '核对报告并补齐所需配置', 'evidence': row.get('evidence', ''), 'source_refs': refs})
+    return report, overview