@midscene/test 1.17.6 → 1.17.7

package/dist/templates/test/android/README.md

@@ -16,6 +16,6 @@ Run tests with `{{TEST_COMMAND}}`. Read `midscene-node-spec.android.md` for the
 
 After changing Node registrations in `midscene.config.ts`, run `{{RUN_NODES_COMMAND}}` to refresh the spec. This loads the configuration without connecting to a device or running tests. Extension factories must only acquire runtime resources inside Node execution.
 
-Tests run serially and stop after the first failed case (`test.bail: 1`). Reports are saved and connections are released on failure. Reports are written to `midscene_run/report/`.
+Tests run serially and stop after the first failed case (`test.bail: 1`). Reports are saved and connections are released on failure. Reports are written to `midscene_run/report/`; structured results and diagnostic records are saved in `midscene_run/test-results/<runId>/`. Set `MIDSCENE_RUN_DIR` to change the artifact root.
 
 Each test file creates its own Agent through `fileSetup` and releases it afterward. The Project is limited to one file at a time because this starter connects to a single device or desktop. Before increasing the Project limit, provide independent devices/displays through your resource allocator.

package/dist/templates/test/android/README.zh.md

@@ -8,6 +8,6 @@
 
 连接设备并用 `adb devices` 确认；可设置 `ANDROID_DEVICE_ID` 指定设备。`cases/todo.yaml` 打开在线 React TodoMVC，添加三个任务、删除一个、完成另一个、筛选已完成任务，并校验可见任务和未完成数量。设备需要能访问该网站；`beforeEach` 先执行 `midscene.config.ts` 中的 `todo.setup`，打开 All 视图并等待输入框加载，再由 `aiAssert` 确认列表为空（包含已完成任务）且没有遮挡弹窗。已有任务会保留；检查失败时，用例会在修改任务前停止。请切换到 All，删除全部任务（包括已完成任务），或仅清除 TodoMVC 网站的数据；关闭遮挡弹窗后重新运行。模型或连接异常应先按报告中的原因排查。AI 检查可能误判，失败时请结合报告确认页面状态。
 
-执行 `{{TEST_COMMAND}}` 运行测试。Node 列表见 `midscene-node-spec.android.md`；修改配置中的 Node 后执行 `{{RUN_NODES_COMMAND}}` 刷新。用例串行运行，首次失败即停止后续用例（`test.bail: 1`）；失败时仍会保存报告并释放连接。报告位于 `midscene_run/report/`。
+执行 `{{TEST_COMMAND}}` 运行测试。Node 列表见 `midscene-node-spec.android.md`；修改配置中的 Node 后执行 `{{RUN_NODES_COMMAND}}` 刷新。用例串行运行，首次失败即停止后续用例（`test.bail: 1`）；失败时仍会保存报告并释放连接。报告位于 `midscene_run/report/`，结构化结果与诊断记录位于 `midscene_run/test-results/<runId>/`。可通过 `MIDSCENE_RUN_DIR` 修改产物根。
 
 每份测试文件通过 `fileSetup` 创建独立 Agent，结束后释放。此模板连接一台设备或一个桌面，因此 Project 的 `maxConcurrentFiles` 固定为 1；提高此上限前，需要通过资源分配脚本为并发文件提供独立设备或桌面。

package/dist/templates/test/android/gitignore

@@ -1,4 +1,3 @@
 node_modules/
 .env
 midscene_run/
-.midscene/test-results/

package/dist/templates/test/computer/README.md

@@ -16,6 +16,6 @@ Run tests with `{{TEST_COMMAND}}`. Read `midscene-node-spec.computer.md` for the
 
 After changing Node registrations in `midscene.config.ts`, run `{{RUN_NODES_COMMAND}}` to refresh the spec. This loads the configuration without connecting to a device or running tests. Extension factories must only acquire runtime resources inside Node execution.
 
-Tests run serially and stop after the first failed case (`test.bail: 1`). Reports are saved and connections are released on failure. Reports are written to `midscene_run/report/`.
+Tests run serially and stop after the first failed case (`test.bail: 1`). Reports are saved and connections are released on failure. Reports are written to `midscene_run/report/`; structured results and diagnostic records are saved in `midscene_run/test-results/<runId>/`. Set `MIDSCENE_RUN_DIR` to change the artifact root.
 
 Each test file creates its own Agent through `fileSetup` and releases it afterward. The Project is limited to one file at a time because this starter connects to a single device or desktop. Before increasing the Project limit, provide independent devices/displays through your resource allocator.

package/dist/templates/test/computer/README.zh.md

@@ -8,6 +8,6 @@
 
 按照[桌面环境指南](https://midscenejs.com/platforms/desktop.html)准备权限和依赖。可设置 `COMPUTER_DISPLAY_ID` 选择显示器；无头 Linux 需要 Xvfb 和 `MIDSCENE_COMPUTER_HEADLESS_LINUX`。`cases/todo.yaml` 在 macOS 使用 Safari、在 Windows/Linux 使用 Chrome 打开在线 React TodoMVC，添加三个任务、删除一个、完成另一个、筛选已完成任务，并校验任务名称及输入框占位文字。桌面需要能访问该网站；`beforeEach` 先执行 `midscene.config.ts` 中的 `todo.setup`，打开 All 视图并等待输入框加载，再由 `aiAssert` 确认列表为空（包含已完成任务）且没有遮挡弹窗。已有任务会保留；检查失败时，用例会在修改任务前停止。请切换到 All，删除全部任务（包括已完成任务），或仅清除 TodoMVC 网站的数据；关闭遮挡弹窗后重新运行。模型或连接异常应先按报告中的原因排查。AI 检查可能误判，失败时请结合报告确认页面状态。
 
-执行 `{{TEST_COMMAND}}` 运行测试。Node 列表见 `midscene-node-spec.computer.md`；修改配置中的 Node 后执行 `{{RUN_NODES_COMMAND}}` 刷新。用例串行运行，首次失败即停止后续用例（`test.bail: 1`）；失败时仍会保存报告并释放连接。报告位于 `midscene_run/report/`。
+执行 `{{TEST_COMMAND}}` 运行测试。Node 列表见 `midscene-node-spec.computer.md`；修改配置中的 Node 后执行 `{{RUN_NODES_COMMAND}}` 刷新。用例串行运行，首次失败即停止后续用例（`test.bail: 1`）；失败时仍会保存报告并释放连接。报告位于 `midscene_run/report/`，结构化结果与诊断记录位于 `midscene_run/test-results/<runId>/`。可通过 `MIDSCENE_RUN_DIR` 修改产物根。
 
 每份测试文件通过 `fileSetup` 创建独立 Agent，结束后释放。此模板连接一台设备或一个桌面，因此 Project 的 `maxConcurrentFiles` 固定为 1；提高此上限前，需要通过资源分配脚本为并发文件提供独立设备或桌面。

package/dist/templates/test/computer/gitignore

@@ -1,4 +1,3 @@
 node_modules/
 .env
 midscene_run/
-.midscene/test-results/

package/dist/templates/test/harmony/README.md

@@ -16,6 +16,6 @@ Run tests with `{{TEST_COMMAND}}`. Read `midscene-node-spec.harmony.md` for the
 
 After changing Node registrations in `midscene.config.ts`, run `{{RUN_NODES_COMMAND}}` to refresh the spec. This loads the configuration without connecting to a device or running tests. Extension factories must only acquire runtime resources inside Node execution.
 
-Tests run serially and stop after the first failed case (`test.bail: 1`). Reports are saved and connections are released on failure. Reports are written to `midscene_run/report/`.
+Tests run serially and stop after the first failed case (`test.bail: 1`). Reports are saved and connections are released on failure. Reports are written to `midscene_run/report/`; structured results and diagnostic records are saved in `midscene_run/test-results/<runId>/`. Set `MIDSCENE_RUN_DIR` to change the artifact root.
 
 Each test file creates its own Agent through `fileSetup` and releases it afterward. The Project is limited to one file at a time because this starter connects to a single device or desktop. Before increasing the Project limit, provide independent devices/displays through your resource allocator.

package/dist/templates/test/harmony/README.zh.md

@@ -8,6 +8,6 @@
 
 连接设备并用 `hdc list targets` 确认；可设置 `HARMONY_DEVICE_ID` 指定设备，必要时设置 `HDC_HOME`。`cases/todo.yaml` 在浏览器打开在线 React TodoMVC，添加三个任务、删除一个、完成另一个、筛选已完成任务，并校验可见任务和未完成数量。设备需要能访问该网站；`beforeEach` 先执行 `midscene.config.ts` 中的 `todo.setup`，打开 All 视图并等待输入框加载，再由 `aiAssert` 确认列表为空（包含已完成任务）且没有遮挡弹窗。已有任务会保留；检查失败时，用例会在修改任务前停止。请切换到 All，删除全部任务（包括已完成任务），或仅清除 TodoMVC 网站的数据；关闭遮挡弹窗后重新运行。模型或连接异常应先按报告中的原因排查。AI 检查可能误判，失败时请结合报告确认页面状态。
 
-执行 `{{TEST_COMMAND}}` 运行测试。Node 列表见 `midscene-node-spec.harmony.md`；修改配置中的 Node 后执行 `{{RUN_NODES_COMMAND}}` 刷新。用例串行运行，首次失败即停止后续用例（`test.bail: 1`）；失败时仍会保存报告并释放连接。报告位于 `midscene_run/report/`。
+执行 `{{TEST_COMMAND}}` 运行测试。Node 列表见 `midscene-node-spec.harmony.md`；修改配置中的 Node 后执行 `{{RUN_NODES_COMMAND}}` 刷新。用例串行运行，首次失败即停止后续用例（`test.bail: 1`）；失败时仍会保存报告并释放连接。报告位于 `midscene_run/report/`，结构化结果与诊断记录位于 `midscene_run/test-results/<runId>/`。可通过 `MIDSCENE_RUN_DIR` 修改产物根。
 
 每份测试文件通过 `fileSetup` 创建独立 Agent，结束后释放。此模板连接一台设备或一个桌面，因此 Project 的 `maxConcurrentFiles` 固定为 1；提高此上限前，需要通过资源分配脚本为并发文件提供独立设备或桌面。

package/dist/templates/test/harmony/gitignore

@@ -1,4 +1,3 @@
 node_modules/
 .env
 midscene_run/
-.midscene/test-results/

package/dist/templates/test/ios/README.md

@@ -16,6 +16,6 @@ Run tests with `{{TEST_COMMAND}}`. Read `midscene-node-spec.ios.md` for the avai
 
 After changing Node registrations in `midscene.config.ts`, run `{{RUN_NODES_COMMAND}}` to refresh the spec. This loads the configuration without connecting to a device or running tests. Extension factories must only acquire runtime resources inside Node execution.
 
-Tests run serially and stop after the first failed case (`test.bail: 1`). Reports are saved and connections are released on failure. Reports are written to `midscene_run/report/`.
+Tests run serially and stop after the first failed case (`test.bail: 1`). Reports are saved and connections are released on failure. Reports are written to `midscene_run/report/`; structured results and diagnostic records are saved in `midscene_run/test-results/<runId>/`. Set `MIDSCENE_RUN_DIR` to change the artifact root.
 
 Each test file creates its own Agent through `fileSetup` and releases it afterward. The Project is limited to one file at a time because this starter connects to a single device or desktop. Before increasing the Project limit, provide independent devices/displays through your resource allocator.

package/dist/templates/test/ios/README.zh.md

@@ -8,6 +8,6 @@
 
 启动 WebDriverAgent。可设置 `WDA_HOST` 和 `WDA_PORT`，或用 `WDA_BASE_URL` 指定带路径前缀的网关地址。远程 MJPEG 流可设置 `WDA_MJPEG_URL`，直连则可设置 `WDA_MJPEG_PORT`。`cases/todo.yaml` 打开在线 React TodoMVC，添加三个任务、删除一个、完成另一个、筛选已完成任务，并校验可见任务和未完成数量。设备需要能访问该网站；`beforeEach` 先执行 `midscene.config.ts` 中的 `todo.setup`，打开 All 视图并等待输入框加载，再由 `aiAssert` 确认列表为空（包含已完成任务）且没有遮挡弹窗。已有任务会保留；检查失败时，用例会在修改任务前停止。请切换到 All，删除全部任务（包括已完成任务），或仅清除 TodoMVC 网站的数据；关闭遮挡弹窗后重新运行。模型或连接异常应先按报告中的原因排查。AI 检查可能误判，失败时请结合报告确认页面状态。
 
-执行 `{{TEST_COMMAND}}` 运行测试。Node 列表见 `midscene-node-spec.ios.md`；修改配置中的 Node 后执行 `{{RUN_NODES_COMMAND}}` 刷新。用例串行运行，首次失败即停止后续用例（`test.bail: 1`）；失败时仍会保存报告并释放连接。报告位于 `midscene_run/report/`。
+执行 `{{TEST_COMMAND}}` 运行测试。Node 列表见 `midscene-node-spec.ios.md`；修改配置中的 Node 后执行 `{{RUN_NODES_COMMAND}}` 刷新。用例串行运行，首次失败即停止后续用例（`test.bail: 1`）；失败时仍会保存报告并释放连接。报告位于 `midscene_run/report/`，结构化结果与诊断记录位于 `midscene_run/test-results/<runId>/`。可通过 `MIDSCENE_RUN_DIR` 修改产物根。
 
 每份测试文件通过 `fileSetup` 创建独立 Agent，结束后释放。此模板连接一台设备或一个桌面，因此 Project 的 `maxConcurrentFiles` 固定为 1；提高此上限前，需要通过资源分配脚本为并发文件提供独立设备或桌面。

package/dist/templates/test/ios/gitignore

@@ -1,4 +1,3 @@
 node_modules/
 .env
 midscene_run/
-.midscene/test-results/

package/dist/templates/test/web/README.md

@@ -16,7 +16,7 @@ Run tests with `{{TEST_COMMAND}}`. Read `midscene-node-spec.web.md` for the avai
 
 After changing Node registrations in `midscene.config.ts`, run `{{RUN_NODES_COMMAND}}` to refresh the spec. This loads the configuration without connecting to a device or running tests. Extension factories must only acquire runtime resources inside Node execution.
 
-Tests run serially and stop after the first failed case (`test.bail: 1`). Reports are saved and the browser is closed on failure. Reports are written to `midscene_run/report/`.
+Tests run serially and stop after the first failed case (`test.bail: 1`). Reports are saved and the browser is closed on failure. Reports are written to `midscene_run/report/`; structured results and diagnostic records are saved in `midscene_run/test-results/<runId>/`. Set `MIDSCENE_RUN_DIR` to change the artifact root.
 
 ## How preparation works
 

package/dist/templates/test/web/README.zh.md

@@ -10,7 +10,7 @@
 
 `cases/todo.yaml` 使用在线 React TodoMVC。每条用例先由 `todo.setup` 打开 All 视图、清空页面存储并重新加载，通过 DOM 确认列表为空，再由 `todo.seed` 添加三个任务；第一条用例校验初始化状态，第二条用 AI 删除和完成任务，再校验最终状态。`todo.captureState` 在每条用例后记录页面状态。运行时需要能访问 TodoMVC 网站。
 
-执行 `{{TEST_COMMAND}}` 运行测试。Node 列表见 `midscene-node-spec.web.md`；修改配置中的 Node 后执行 `{{RUN_NODES_COMMAND}}` 刷新。用例串行运行，首次失败即停止后续用例（`test.bail: 1`）；失败时仍会保存报告并关闭浏览器。报告位于 `midscene_run/report/`。
+执行 `{{TEST_COMMAND}}` 运行测试。Node 列表见 `midscene-node-spec.web.md`；修改配置中的 Node 后执行 `{{RUN_NODES_COMMAND}}` 刷新。用例串行运行，首次失败即停止后续用例（`test.bail: 1`）；失败时仍会保存报告并关闭浏览器。报告位于 `midscene_run/report/`，结构化结果与诊断记录位于 `midscene_run/test-results/<runId>/`。可通过 `MIDSCENE_RUN_DIR` 修改产物根。
 
 ## 准备工作如何执行
 

package/dist/templates/test/web/gitignore

@@ -1,4 +1,3 @@
 node_modules/
 .env
 midscene_run/
-.midscene/test-results/

package/dist/types/cli/index.d.ts

@@ -151,7 +151,7 @@ export declare interface ResolvedTestOptions {
 }
 
 export declare interface ResolvedTestOutputDefinition {
-    reportDir: string;
+    reportDir?: string;
     reportFormat: 'single-html' | 'html-and-external-assets';
 }
 
@@ -194,6 +194,7 @@ export declare interface TestOptions {
 }
 
 export declare interface TestOutputDefinition {
+    /** Explicit report location, relative to the Test project. Defaults to report/ under the artifact root. */
     reportDir?: string;
     /** Defaults to a self-contained HTML; external assets require HTTP preview. */
     reportFormat?: 'single-html' | 'html-and-external-assets';

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@midscene/test",
   "description": "Midscene Test: an extensible YAML testing framework and Node SDK.",
-  "version": "1.17.6",
+  "version": "1.17.7",
   "repository": {
     "type": "git",
     "url": "https://github.com/web-infra-dev/midscene.git",
@@ -59,8 +59,8 @@
     "tsx": "^4.19.2",
     "yargs": "17.7.3",
     "zod": "^3.25.1",
-    "@midscene/core": "1.17.6",
-    "@midscene/shared": "1.17.6"
+    "@midscene/shared": "1.17.7",
+    "@midscene/core": "1.17.7"
   },
   "devDependencies": {
     "@rslib/core": "^0.18.3",
@@ -72,20 +72,20 @@
     "puppeteer": "24.6.0",
     "typescript": "^5.8.3",
     "vitest": "3.0.5",
-    "@midscene/android": "1.17.6",
-    "@midscene/computer": "1.17.6",
-    "@midscene/ios": "1.17.6",
-    "@midscene/harmony": "1.17.6",
-    "@midscene/web": "1.17.6"
+    "@midscene/android": "1.17.7",
+    "@midscene/harmony": "1.17.7",
+    "@midscene/ios": "1.17.7",
+    "@midscene/web": "1.17.7",
+    "@midscene/computer": "1.17.7"
   },
   "peerDependencies": {
     "playwright": "^1.45.0",
     "puppeteer": "24.6.0",
-    "@midscene/android": "1.17.6",
-    "@midscene/computer": "1.17.6",
-    "@midscene/ios": "1.17.6",
-    "@midscene/web": "1.17.6",
-    "@midscene/harmony": "1.17.6"
+    "@midscene/android": "1.17.7",
+    "@midscene/harmony": "1.17.7",
+    "@midscene/ios": "1.17.7",
+    "@midscene/computer": "1.17.7",
+    "@midscene/web": "1.17.7"
   },
   "peerDependenciesMeta": {
     "@midscene/android": {