@midscene/core 0.27.4-beta-20250826121712.0 → 0.27.5-beta-20250827031507.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (45) hide show
  1. package/dist/es/agent/agent.mjs +25 -22
  2. package/dist/es/agent/agent.mjs.map +1 -1
  3. package/dist/es/agent/index.mjs +2 -2
  4. package/dist/es/agent/tasks.mjs +47 -44
  5. package/dist/es/agent/tasks.mjs.map +1 -1
  6. package/dist/es/agent/utils.mjs +17 -17
  7. package/dist/es/agent/utils.mjs.map +1 -1
  8. package/dist/es/ai-model/common.mjs +1 -1
  9. package/dist/es/ai-model/common.mjs.map +1 -1
  10. package/dist/es/ai-model/llm-planning.mjs.map +1 -1
  11. package/dist/es/device/index.mjs +2 -2
  12. package/dist/es/device/index.mjs.map +1 -1
  13. package/dist/es/index.mjs.map +1 -1
  14. package/dist/es/types.mjs.map +1 -1
  15. package/dist/es/utils.mjs +2 -2
  16. package/dist/es/yaml/player.mjs +6 -6
  17. package/dist/es/yaml/player.mjs.map +1 -1
  18. package/dist/lib/agent/agent.js +24 -21
  19. package/dist/lib/agent/agent.js.map +1 -1
  20. package/dist/lib/agent/index.js +3 -3
  21. package/dist/lib/agent/tasks.js +49 -46
  22. package/dist/lib/agent/tasks.js.map +1 -1
  23. package/dist/lib/agent/utils.js +17 -17
  24. package/dist/lib/agent/utils.js.map +1 -1
  25. package/dist/lib/ai-model/common.js +1 -1
  26. package/dist/lib/ai-model/common.js.map +1 -1
  27. package/dist/lib/ai-model/llm-planning.js.map +1 -1
  28. package/dist/lib/device/index.js +4 -4
  29. package/dist/lib/device/index.js.map +1 -1
  30. package/dist/lib/index.js.map +1 -1
  31. package/dist/lib/types.js.map +1 -1
  32. package/dist/lib/utils.js +2 -2
  33. package/dist/lib/yaml/player.js +6 -6
  34. package/dist/lib/yaml/player.js.map +1 -1
  35. package/dist/types/agent/agent.d.ts +10 -9
  36. package/dist/types/agent/index.d.ts +2 -2
  37. package/dist/types/agent/tasks.d.ts +5 -4
  38. package/dist/types/agent/utils.d.ts +5 -5
  39. package/dist/types/ai-model/llm-planning.d.ts +2 -2
  40. package/dist/types/device/index.d.ts +3 -2
  41. package/dist/types/index.d.ts +1 -1
  42. package/dist/types/types.d.ts +2 -3
  43. package/dist/types/yaml/player.d.ts +1 -1
  44. package/dist/types/yaml.d.ts +1 -1
  45. package/package.json +3 -3
@@ -6,7 +6,7 @@ import { MIDSCENE_CACHE, globalConfigManager, vlLocateMode } from "@midscene/sha
6
6
  import { getDebug } from "@midscene/shared/logger";
7
7
  import { assert } from "@midscene/shared/utils";
8
8
  import { TaskCache } from "./task-cache.mjs";
9
- import { PageTaskExecutor, locatePlanForLocate } from "./tasks.mjs";
9
+ import { TaskExecutor, locatePlanForLocate } from "./tasks.mjs";
10
10
  import { locateParamStr, paramStr, taskTitleStr, typeStr } from "./ui-utils.mjs";
11
11
  import { commonContextParser, getReportFileName, parsePrompt, printReportMsg, trimContextByViewport } from "./utils.mjs";
12
12
  function _define_property(obj, key, value) {
@@ -35,20 +35,23 @@ const defaultInsightExtractOption = {
35
35
  screenshotIncluded: true
36
36
  };
37
37
  class Agent {
38
+ get page() {
39
+ return this.interface;
40
+ }
38
41
  async getActionSpace() {
39
- return this.page.actionSpace();
42
+ return this.interface.actionSpace();
40
43
  }
41
44
  async getUIContext(action) {
42
- if (this.frozenPageContext) {
45
+ if (this.frozenUIContext) {
43
46
  debug('Using frozen page context for action:', action);
44
- return this.frozenPageContext;
47
+ return this.frozenUIContext;
45
48
  }
46
- if (this.page.getContext) {
49
+ if (this.interface.getContext) {
47
50
  debug('Using page.getContext for action:', action);
48
- return await this.page.getContext();
51
+ return await this.interface.getContext();
49
52
  }
50
53
  debug('Using commonContextParser for action:', action);
51
- return await commonContextParser(this.page);
54
+ return await commonContextParser(this.interface);
52
55
  }
53
56
  async _snapshotContext() {
54
57
  return await this.getUIContext('locate');
@@ -335,7 +338,7 @@ class Agent {
335
338
  return {
336
339
  rect: null == element ? void 0 : element.rect,
337
340
  center: null == element ? void 0 : element.center,
338
- scale: (await this.page.size()).dpr
341
+ scale: (await this.interface.size()).dpr
339
342
  };
340
343
  }
341
344
  async aiAssert(assertion, msg, opt) {
@@ -395,16 +398,16 @@ class Agent {
395
398
  };
396
399
  }
397
400
  async evaluateJavaScript(script) {
398
- assert(this.page.evaluateJavaScript, 'evaluateJavaScript is not supported in current agent');
399
- return this.page.evaluateJavaScript(script);
401
+ assert(this.interface.evaluateJavaScript, 'evaluateJavaScript is not supported in current agent');
402
+ return this.interface.evaluateJavaScript(script);
400
403
  }
401
404
  async destroy() {
402
- await this.page.destroy();
405
+ await this.interface.destroy();
403
406
  this.resetDump();
404
407
  this.destroyed = true;
405
408
  }
406
409
  async logScreenshot(title, opt) {
407
- const base64 = await this.page.screenshotBase64();
410
+ const base64 = await this.interface.screenshotBase64();
408
411
  const now = Date.now();
409
412
  const recorder = [
410
413
  {
@@ -452,7 +455,7 @@ class Agent {
452
455
  const { tasks, ...restExecution } = execution;
453
456
  let newTasks = tasks;
454
457
  if (Array.isArray(tasks)) newTasks = tasks.map((task)=>{
455
- const { pageContext, log, ...restTask } = task;
458
+ const { uiContext, log, ...restTask } = task;
456
459
  return restTask;
457
460
  });
458
461
  return {
@@ -472,16 +475,16 @@ class Agent {
472
475
  debug('Freezing page context');
473
476
  const context = await this._snapshotContext();
474
477
  context._isFrozen = true;
475
- this.frozenPageContext = context;
478
+ this.frozenUIContext = context;
476
479
  debug('Page context frozen successfully');
477
480
  }
478
481
  async unfreezePageContext() {
479
482
  debug('Unfreezing page context');
480
- this.frozenPageContext = void 0;
483
+ this.frozenUIContext = void 0;
481
484
  debug('Page context unfrozen successfully');
482
485
  }
483
- constructor(page, opts){
484
- _define_property(this, "page", void 0);
486
+ constructor(interfaceInstance, opts){
487
+ _define_property(this, "interface", void 0);
485
488
  _define_property(this, "insight", void 0);
486
489
  _define_property(this, "dump", void 0);
487
490
  _define_property(this, "reportFile", void 0);
@@ -493,8 +496,8 @@ class Agent {
493
496
  _define_property(this, "taskCache", void 0);
494
497
  _define_property(this, "onDumpUpdate", void 0);
495
498
  _define_property(this, "destroyed", false);
496
- _define_property(this, "frozenPageContext", void 0);
497
- this.page = page;
499
+ _define_property(this, "frozenUIContext", void 0);
500
+ this.interface = interfaceInstance;
498
501
  this.opts = Object.assign({
499
502
  generateReport: true,
500
503
  autoPrintReportMsg: true,
@@ -505,13 +508,13 @@ class Agent {
505
508
  globalConfigManager.init(null == opts ? void 0 : opts.modelConfig);
506
509
  this.onTaskStartTip = this.opts.onTaskStartTip;
507
510
  this.insight = new Insight(async (action)=>this.getUIContext(action));
508
- if ((null == opts ? void 0 : opts.cacheId) && 'android' !== this.page.pageType) this.taskCache = new TaskCache(opts.cacheId, globalConfigManager.getEnvConfigInBoolean(MIDSCENE_CACHE));
509
- this.taskExecutor = new PageTaskExecutor(this.page, this.insight, {
511
+ if ((null == opts ? void 0 : opts.cacheId) && 'android' !== this.interface.interfaceType) this.taskCache = new TaskCache(opts.cacheId, globalConfigManager.getEnvConfigInBoolean(MIDSCENE_CACHE));
512
+ this.taskExecutor = new TaskExecutor(this.interface, this.insight, {
510
513
  taskCache: this.taskCache,
511
514
  onTaskStart: this.callbackOnTaskStartTip.bind(this)
512
515
  });
513
516
  this.dump = this.resetDump();
514
- this.reportFileName = (null == opts ? void 0 : opts.reportFileName) || getReportFileName((null == opts ? void 0 : opts.testId) || this.page.pageType || 'web');
517
+ this.reportFileName = (null == opts ? void 0 : opts.reportFileName) || getReportFileName((null == opts ? void 0 : opts.testId) || this.interface.interfaceType || 'web');
515
518
  }
516
519
  }
517
520
  export { Agent };
@@ -1 +1 @@
1
- {"version":3,"file":"agent/agent.mjs","sources":["webpack://@midscene/core/./src/agent/agent.ts"],"sourcesContent":["import {\n type AgentAssertOpt,\n type AgentDescribeElementAtPointResult,\n type AgentWaitForOpt,\n type DeviceAction,\n type ExecutionDump,\n type ExecutionRecorderItem,\n type ExecutionTask,\n type ExecutionTaskLog,\n type Executor,\n type GroupedActionDump,\n Insight,\n type InsightAction,\n type InsightExtractOption,\n type InsightExtractParam,\n type LocateOption,\n type LocateResultElement,\n type LocateValidatorResult,\n type LocatorValidatorOption,\n type MidsceneYamlScript,\n type OnTaskStartTip,\n type PlanningAction,\n type Rect,\n type ScrollParam,\n type TUserPrompt,\n type UIContext,\n} from '../index';\n\nimport yaml from 'js-yaml';\n\nimport {\n groupedActionDumpFileExt,\n reportHTMLContent,\n stringifyDumpData,\n writeLogFile,\n} from '@/utils';\nimport {\n ScriptPlayer,\n buildDetailedLocateParam,\n parseYamlScript,\n} from '../yaml/index';\n\nimport type { AbstractDevice } from '@/device';\nimport {\n type IModelPreferences,\n MIDSCENE_CACHE,\n type TModelConfigFn,\n globalConfigManager,\n vlLocateMode,\n} from '@midscene/shared/env';\nimport { getDebug } from '@midscene/shared/logger';\nimport { assert } from '@midscene/shared/utils';\n// import type { AndroidDeviceInputOpt } from '../device';\nimport { TaskCache } from './task-cache';\nimport { PageTaskExecutor, locatePlanForLocate } from './tasks';\nimport { locateParamStr, paramStr, taskTitleStr, typeStr } from './ui-utils';\nimport {\n commonContextParser,\n getReportFileName,\n parsePrompt,\n printReportMsg,\n} from './utils';\nimport { trimContextByViewport } from './utils';\n\nconst debug = getDebug('agent');\n\nconst distanceOfTwoPoints = (p1: [number, number], p2: [number, number]) => {\n const [x1, y1] = p1;\n const [x2, y2] = p2;\n return Math.round(Math.sqrt((x1 - x2) ** 2 + (y1 - y2) ** 2));\n};\n\nconst includedInRect = (point: [number, number], rect: Rect) => {\n const [x, y] = point;\n const { left, top, width, height } = rect;\n return x >= left && x <= left + width && y >= top && y <= top + height;\n};\n\nconst defaultInsightExtractOption: InsightExtractOption = {\n domIncluded: false,\n screenshotIncluded: true,\n};\n\nexport interface PageAgentOpt {\n testId?: string;\n cacheId?: string;\n groupName?: string;\n groupDescription?: string;\n /* if auto generate report, default true */\n generateReport?: boolean;\n /* if auto print report msg, default true */\n autoPrintReportMsg?: boolean;\n onTaskStartTip?: OnTaskStartTip;\n aiActionContext?: string;\n /* custom report file name */\n reportFileName?: string;\n modelConfig?: TModelConfigFn;\n}\n\nexport class Agent<PageType extends AbstractDevice = AbstractDevice> {\n page: PageType;\n\n insight: Insight;\n\n dump: GroupedActionDump;\n\n reportFile?: string | null;\n\n reportFileName?: string;\n\n taskExecutor: PageTaskExecutor;\n\n opts: PageAgentOpt;\n\n /**\n * If true, the agent will not perform any actions\n */\n dryMode = false;\n\n onTaskStartTip?: OnTaskStartTip;\n\n taskCache?: TaskCache;\n\n onDumpUpdate?: (dump: string) => void;\n\n destroyed = false;\n\n /**\n * Frozen page context for consistent AI operations\n */\n private frozenPageContext?: UIContext;\n\n constructor(page: PageType, opts?: PageAgentOpt) {\n this.page = page;\n this.opts = Object.assign(\n {\n generateReport: true,\n autoPrintReportMsg: true,\n groupName: 'Midscene Report',\n groupDescription: '',\n },\n opts || {},\n );\n if (opts?.modelConfig && typeof opts?.modelConfig !== 'function') {\n throw new Error(\n `opts.modelConfig must be one of function or undefined, but got ${typeof opts?.modelConfig}`,\n );\n }\n\n globalConfigManager.init(opts?.modelConfig);\n\n this.onTaskStartTip = this.opts.onTaskStartTip;\n // get the parent browser of the puppeteer page\n // const browser = (this.page as PuppeteerWebPage).browser();\n\n this.insight = new Insight(async (action: InsightAction) => {\n return this.getUIContext(action);\n });\n\n if (opts?.cacheId && this.page.pageType !== 'android') {\n this.taskCache = new TaskCache(\n opts.cacheId,\n globalConfigManager.getEnvConfigInBoolean(MIDSCENE_CACHE), // if we should use cache to match the element\n );\n }\n\n this.taskExecutor = new PageTaskExecutor(this.page, this.insight, {\n taskCache: this.taskCache,\n onTaskStart: this.callbackOnTaskStartTip.bind(this),\n });\n this.dump = this.resetDump();\n this.reportFileName =\n opts?.reportFileName ||\n getReportFileName(opts?.testId || this.page.pageType || 'web');\n }\n\n async getActionSpace(): Promise<DeviceAction[]> {\n return this.page.actionSpace();\n }\n\n async getUIContext(action?: InsightAction): Promise<UIContext> {\n // If page context is frozen, return the frozen context for all actions\n if (this.frozenPageContext) {\n debug('Using frozen page context for action:', action);\n return this.frozenPageContext;\n }\n\n if (this.page.getContext) {\n debug('Using page.getContext for action:', action);\n return await this.page.getContext();\n } else {\n debug('Using commonContextParser for action:', action);\n return await commonContextParser(this.page);\n }\n }\n\n async _snapshotContext(): Promise<UIContext> {\n return await this.getUIContext('locate');\n }\n\n async setAIActionContext(prompt: string) {\n this.opts.aiActionContext = prompt;\n }\n\n resetDump() {\n this.dump = {\n groupName: this.opts.groupName!,\n groupDescription: this.opts.groupDescription,\n executions: [],\n modelBriefs: [],\n };\n\n return this.dump;\n }\n\n appendExecutionDump(execution: ExecutionDump) {\n // use trimContextByViewport to process execution\n const trimmedExecution = trimContextByViewport(execution);\n const currentDump = this.dump;\n currentDump.executions.push(trimmedExecution);\n }\n\n dumpDataString() {\n // update dump info\n this.dump.groupName = this.opts.groupName!;\n this.dump.groupDescription = this.opts.groupDescription;\n return stringifyDumpData(this.dump);\n }\n\n reportHTMLString() {\n return reportHTMLContent(this.dumpDataString());\n }\n\n writeOutActionDumps() {\n if (this.destroyed) {\n throw new Error(\n 'PageAgent has been destroyed. Cannot update report file.',\n );\n }\n const { generateReport, autoPrintReportMsg } = this.opts;\n this.reportFile = writeLogFile({\n fileName: this.reportFileName!,\n fileExt: groupedActionDumpFileExt,\n fileContent: this.dumpDataString(),\n type: 'dump',\n generateReport,\n });\n debug('writeOutActionDumps', this.reportFile);\n if (generateReport && autoPrintReportMsg && this.reportFile) {\n printReportMsg(this.reportFile);\n }\n }\n\n private async callbackOnTaskStartTip(task: ExecutionTask) {\n const param = paramStr(task);\n const tip = param ? `${typeStr(task)} - ${param}` : typeStr(task);\n\n if (this.onTaskStartTip) {\n await this.onTaskStartTip(tip);\n }\n }\n\n private async afterTaskRunning(executor: Executor, doNotThrowError = false) {\n this.appendExecutionDump(executor.dump());\n\n try {\n await this.onDumpUpdate?.(this.dumpDataString());\n } catch (error) {\n console.error('Error in onDumpUpdate', error);\n }\n\n this.writeOutActionDumps();\n\n if (executor.isInErrorState() && !doNotThrowError) {\n const errorTask = executor.latestErrorTask();\n throw new Error(`${errorTask?.errorMessage}\\n${errorTask?.errorStack}`, {\n cause: errorTask?.error,\n });\n }\n }\n\n async callActionInActionSpace<T = any>(\n type: string,\n opt?: T, // and all other action params\n ) {\n debug('callActionInActionSpace', type, ',', opt, ',', opt);\n\n const actionPlan: PlanningAction<T> = {\n type: type as any,\n param: (opt as any) || {},\n thought: '',\n };\n debug('actionPlan', actionPlan); // , ', in which the locateParam is', locateParam);\n\n const plans: PlanningAction[] = [actionPlan].filter(\n Boolean,\n ) as PlanningAction[];\n\n const title = taskTitleStr(\n type as any,\n locateParamStr((opt as any)?.locate || {}),\n );\n\n const { output, executor } = await this.taskExecutor.runPlans(title, plans);\n await this.afterTaskRunning(executor);\n return output;\n }\n\n async aiTap(locatePrompt: TUserPrompt, opt?: LocateOption) {\n assert(locatePrompt, 'missing locate prompt for tap');\n\n const detailedLocateParam = buildDetailedLocateParam(locatePrompt, opt);\n\n return this.callActionInActionSpace('Tap', {\n locate: detailedLocateParam,\n });\n }\n\n async aiRightClick(locatePrompt: TUserPrompt, opt?: LocateOption) {\n assert(locatePrompt, 'missing locate prompt for right click');\n\n const detailedLocateParam = buildDetailedLocateParam(locatePrompt, opt);\n\n return this.callActionInActionSpace('RightClick', {\n locate: detailedLocateParam,\n });\n }\n\n async aiHover(locatePrompt: TUserPrompt, opt?: LocateOption) {\n assert(locatePrompt, 'missing locate prompt for hover');\n\n const detailedLocateParam = buildDetailedLocateParam(locatePrompt, opt);\n\n return this.callActionInActionSpace('Hover', {\n locate: detailedLocateParam,\n });\n }\n\n // New signature, always use locatePrompt as the first param\n async aiInput(\n locatePrompt: TUserPrompt,\n opt: LocateOption & { value: string }, // AndroidDeviceInputOpt &\n ): Promise<any>;\n\n // Legacy signature - deprecated\n /**\n * @deprecated Use aiInput(locatePrompt, opt) instead where opt contains the value\n */\n async aiInput(\n value: string,\n locatePrompt: TUserPrompt,\n opt?: LocateOption, // AndroidDeviceInputOpt &\n ): Promise<any>;\n\n // Implementation\n async aiInput(\n locatePromptOrValue: TUserPrompt | string,\n locatePromptOrOpt:\n | TUserPrompt\n | (LocateOption & { value: string }) // AndroidDeviceInputOpt &\n | undefined,\n optOrUndefined?: LocateOption, // AndroidDeviceInputOpt &\n ) {\n let value: string;\n let locatePrompt: TUserPrompt;\n let opt:\n | (LocateOption & { value: string }) // AndroidDeviceInputOpt &\n | undefined;\n\n // Check if using new signature (first param is locatePrompt, second has value)\n if (\n typeof locatePromptOrOpt === 'object' &&\n locatePromptOrOpt !== null &&\n 'value' in locatePromptOrOpt\n ) {\n // New signature: aiInput(locatePrompt, opt)\n locatePrompt = locatePromptOrValue as TUserPrompt;\n const optWithValue = locatePromptOrOpt as LocateOption & {\n // AndroidDeviceInputOpt &\n value: string;\n };\n value = optWithValue.value;\n opt = optWithValue;\n } else {\n // Legacy signature: aiInput(value, locatePrompt, opt)\n value = locatePromptOrValue as string;\n locatePrompt = locatePromptOrOpt as TUserPrompt;\n opt = {\n ...optOrUndefined,\n value,\n };\n }\n\n assert(\n typeof value === 'string',\n 'input value must be a string, use empty string if you want to clear the input',\n );\n assert(locatePrompt, 'missing locate prompt for input');\n\n const detailedLocateParam = buildDetailedLocateParam(locatePrompt, opt);\n\n return this.callActionInActionSpace('Input', {\n ...(opt || {}),\n locate: detailedLocateParam,\n });\n }\n\n // New signature\n async aiKeyboardPress(\n locatePrompt: TUserPrompt,\n opt: LocateOption & { keyName: string },\n ): Promise<any>;\n\n // Legacy signature - deprecated\n /**\n * @deprecated Use aiKeyboardPress(locatePrompt, opt) instead where opt contains the keyName\n */\n async aiKeyboardPress(\n keyName: string,\n locatePrompt?: TUserPrompt,\n opt?: LocateOption,\n ): Promise<any>;\n\n // Implementation\n async aiKeyboardPress(\n locatePromptOrKeyName: TUserPrompt | string,\n locatePromptOrOpt:\n | TUserPrompt\n | (LocateOption & { keyName: string })\n | undefined,\n optOrUndefined?: LocateOption,\n ) {\n let keyName: string;\n let locatePrompt: TUserPrompt | undefined;\n let opt: (LocateOption & { keyName: string }) | undefined;\n\n // Check if using new signature (first param is locatePrompt, second has keyName)\n if (\n typeof locatePromptOrOpt === 'object' &&\n locatePromptOrOpt !== null &&\n 'keyName' in locatePromptOrOpt\n ) {\n // New signature: aiKeyboardPress(locatePrompt, opt)\n locatePrompt = locatePromptOrKeyName as TUserPrompt;\n opt = locatePromptOrOpt as LocateOption & {\n keyName: string;\n };\n } else {\n // Legacy signature: aiKeyboardPress(keyName, locatePrompt, opt)\n keyName = locatePromptOrKeyName as string;\n locatePrompt = locatePromptOrOpt as TUserPrompt | undefined;\n opt = {\n ...(optOrUndefined || {}),\n keyName,\n };\n }\n\n assert(opt?.keyName, 'missing keyName for keyboard press');\n\n const detailedLocateParam = locatePrompt\n ? buildDetailedLocateParam(locatePrompt, opt)\n : undefined;\n\n return this.callActionInActionSpace('KeyboardPress', {\n ...(opt || {}),\n locate: detailedLocateParam,\n });\n }\n\n // New signature\n async aiScroll(\n locatePrompt: TUserPrompt | undefined,\n opt: LocateOption & ScrollParam,\n ): Promise<any>;\n\n // Legacy signature - deprecated\n /**\n * @deprecated Use aiScroll(locatePrompt, opt) instead where opt contains the scroll parameters\n */\n async aiScroll(\n scrollParam: ScrollParam,\n locatePrompt?: TUserPrompt,\n opt?: LocateOption,\n ): Promise<any>;\n\n // Implementation\n async aiScroll(\n locatePromptOrScrollParam: TUserPrompt | ScrollParam | undefined,\n locatePromptOrOpt: TUserPrompt | (LocateOption & ScrollParam) | undefined,\n optOrUndefined?: LocateOption,\n ) {\n let scrollParam: ScrollParam;\n let locatePrompt: TUserPrompt | undefined;\n let opt: LocateOption | undefined;\n\n // Check if using new signature (first param is locatePrompt, second has scroll params)\n if (\n typeof locatePromptOrOpt === 'object' &&\n ('direction' in locatePromptOrOpt ||\n 'scrollType' in locatePromptOrOpt ||\n 'distance' in locatePromptOrOpt)\n ) {\n // New signature: aiScroll(locatePrompt, opt)\n locatePrompt = locatePromptOrScrollParam as TUserPrompt;\n opt = locatePromptOrOpt as LocateOption & ScrollParam;\n } else {\n // Legacy signature: aiScroll(scrollParam, locatePrompt, opt)\n scrollParam = locatePromptOrScrollParam as ScrollParam;\n locatePrompt = locatePromptOrOpt as TUserPrompt | undefined;\n opt = {\n ...(optOrUndefined || {}),\n ...(scrollParam || {}),\n };\n }\n\n const detailedLocateParam = buildDetailedLocateParam(\n locatePrompt || '',\n opt,\n );\n\n return this.callActionInActionSpace('Scroll', {\n ...(opt || {}),\n locate: detailedLocateParam,\n });\n }\n\n async aiAction(\n taskPrompt: string,\n opt?: {\n cacheable?: boolean;\n },\n ) {\n const modelPreferences: IModelPreferences = { intent: 'planning' };\n const cacheable = opt?.cacheable;\n // if vlm-ui-tars, plan cache is not used\n const isVlmUiTars = vlLocateMode(modelPreferences) === 'vlm-ui-tars';\n const matchedCache =\n isVlmUiTars || cacheable === false\n ? undefined\n : this.taskCache?.matchPlanCache(taskPrompt);\n if (matchedCache && this.taskCache?.isCacheResultUsed) {\n // log into report file\n const { executor } = await this.taskExecutor.loadYamlFlowAsPlanning(\n taskPrompt,\n matchedCache.cacheContent?.yamlWorkflow,\n );\n\n await await this.afterTaskRunning(executor);\n\n debug('matched cache, will call .runYaml to run the action');\n const yaml = matchedCache.cacheContent?.yamlWorkflow;\n return this.runYaml(yaml);\n }\n\n const { output, executor } = await (isVlmUiTars\n ? this.taskExecutor.actionToGoal(taskPrompt)\n : this.taskExecutor.action(taskPrompt, this.opts.aiActionContext));\n\n // update cache\n if (this.taskCache && output?.yamlFlow && cacheable !== false) {\n const yamlContent: MidsceneYamlScript = {\n tasks: [\n {\n name: taskPrompt,\n flow: output.yamlFlow,\n },\n ],\n };\n const yamlFlowStr = yaml.dump(yamlContent);\n this.taskCache.updateOrAppendCacheRecord(\n {\n type: 'plan',\n prompt: taskPrompt,\n yamlWorkflow: yamlFlowStr,\n },\n matchedCache,\n );\n }\n\n await this.afterTaskRunning(executor);\n return output;\n }\n\n async aiQuery<ReturnType = any>(\n demand: InsightExtractParam,\n opt: InsightExtractOption = defaultInsightExtractOption,\n ): Promise<ReturnType> {\n const { output, executor } =\n await this.taskExecutor.createTypeQueryExecution('Query', demand, opt);\n await this.afterTaskRunning(executor);\n return output as ReturnType;\n }\n\n async aiBoolean(\n prompt: TUserPrompt,\n opt: InsightExtractOption = defaultInsightExtractOption,\n ): Promise<boolean> {\n const { textPrompt, multimodalPrompt } = parsePrompt(prompt);\n const { output, executor } =\n await this.taskExecutor.createTypeQueryExecution(\n 'Boolean',\n textPrompt,\n opt,\n multimodalPrompt,\n );\n await this.afterTaskRunning(executor);\n return output as boolean;\n }\n\n async aiNumber(\n prompt: TUserPrompt,\n opt: InsightExtractOption = defaultInsightExtractOption,\n ): Promise<number> {\n const { textPrompt, multimodalPrompt } = parsePrompt(prompt);\n const { output, executor } =\n await this.taskExecutor.createTypeQueryExecution(\n 'Number',\n textPrompt,\n opt,\n multimodalPrompt,\n );\n await this.afterTaskRunning(executor);\n return output as number;\n }\n\n async aiString(\n prompt: TUserPrompt,\n opt: InsightExtractOption = defaultInsightExtractOption,\n ): Promise<string> {\n const { textPrompt, multimodalPrompt } = parsePrompt(prompt);\n const { output, executor } =\n await this.taskExecutor.createTypeQueryExecution(\n 'String',\n textPrompt,\n opt,\n multimodalPrompt,\n );\n await this.afterTaskRunning(executor);\n return output as string;\n }\n\n async aiAsk(\n prompt: TUserPrompt,\n opt: InsightExtractOption = defaultInsightExtractOption,\n ): Promise<string> {\n return this.aiString(prompt, opt);\n }\n\n async describeElementAtPoint(\n center: [number, number],\n opt?: {\n verifyPrompt?: boolean;\n retryLimit?: number;\n deepThink?: boolean;\n } & LocatorValidatorOption,\n ): Promise<AgentDescribeElementAtPointResult> {\n const { verifyPrompt = true, retryLimit = 3 } = opt || {};\n\n let success = false;\n let retryCount = 0;\n let resultPrompt = '';\n let deepThink = opt?.deepThink || false;\n let verifyResult: LocateValidatorResult | undefined;\n\n while (!success && retryCount < retryLimit) {\n if (retryCount >= 2) {\n deepThink = true;\n }\n debug(\n 'aiDescribe',\n center,\n 'verifyPrompt',\n verifyPrompt,\n 'retryCount',\n retryCount,\n 'deepThink',\n deepThink,\n );\n const text = await this.insight.describe(center, { deepThink });\n debug('aiDescribe text', text);\n assert(text.description, `failed to describe element at [${center}]`);\n resultPrompt = text.description;\n\n verifyResult = await this.verifyLocator(\n resultPrompt,\n deepThink ? { deepThink: true } : undefined,\n center,\n opt,\n );\n if (verifyResult.pass) {\n success = true;\n } else {\n retryCount++;\n }\n }\n\n return {\n prompt: resultPrompt,\n deepThink,\n verifyResult,\n };\n }\n\n async verifyLocator(\n prompt: string,\n locateOpt: LocateOption | undefined,\n expectCenter: [number, number],\n verifyLocateOption?: LocatorValidatorOption,\n ): Promise<LocateValidatorResult> {\n debug('verifyLocator', prompt, locateOpt, expectCenter, verifyLocateOption);\n\n const { center: verifyCenter, rect: verifyRect } = await this.aiLocate(\n prompt,\n locateOpt,\n );\n const distance = distanceOfTwoPoints(expectCenter, verifyCenter);\n const included = includedInRect(expectCenter, verifyRect);\n const pass =\n distance <= (verifyLocateOption?.centerDistanceThreshold || 20) ||\n included;\n const verifyResult = {\n pass,\n rect: verifyRect,\n center: verifyCenter,\n centerDistance: distance,\n };\n debug('aiDescribe verifyResult', verifyResult);\n return verifyResult;\n }\n\n async aiLocate(prompt: TUserPrompt, opt?: LocateOption) {\n const locateParam = buildDetailedLocateParam(prompt, opt);\n assert(locateParam, 'cannot get locate param for aiLocate');\n const locatePlan = locatePlanForLocate(locateParam);\n const plans = [locatePlan];\n const { executor, output } = await this.taskExecutor.runPlans(\n taskTitleStr('Locate', locateParamStr(locateParam)),\n plans,\n );\n await this.afterTaskRunning(executor);\n\n const { element } = output;\n\n return {\n rect: element?.rect,\n center: element?.center,\n scale: (await this.page.size()).dpr,\n } as Pick<LocateResultElement, 'rect' | 'center'> & {\n scale: number;\n };\n }\n\n async aiAssert(\n assertion: TUserPrompt,\n msg?: string,\n opt?: AgentAssertOpt & InsightExtractOption,\n ) {\n const insightOpt: InsightExtractOption = {\n domIncluded: opt?.domIncluded ?? defaultInsightExtractOption.domIncluded,\n screenshotIncluded:\n opt?.screenshotIncluded ??\n defaultInsightExtractOption.screenshotIncluded,\n returnThought: opt?.returnThought ?? true,\n isWaitForAssert: opt?.isWaitForAssert,\n doNotThrowError: opt?.doNotThrowError,\n };\n\n const { output, executor, thought } = await this.taskExecutor.assert(\n assertion,\n insightOpt,\n );\n await this.afterTaskRunning(executor, true);\n\n const message = output\n ? undefined\n : `Assertion failed: ${msg || assertion}\\nReason: ${\n thought || executor.latestErrorTask()?.error || '(no_reason)'\n }`;\n\n if (opt?.keepRawResponse) {\n return {\n pass: output,\n thought,\n message,\n };\n }\n\n if (!output) {\n throw new Error(message);\n }\n }\n\n async aiWaitFor(assertion: TUserPrompt, opt?: AgentWaitForOpt) {\n const { executor } = await this.taskExecutor.waitFor(assertion, {\n timeoutMs: opt?.timeoutMs || 15 * 1000,\n checkIntervalMs: opt?.checkIntervalMs || 3 * 1000,\n });\n await this.afterTaskRunning(executor, true);\n\n if (executor.isInErrorState()) {\n const errorTask = executor.latestErrorTask();\n throw new Error(`${errorTask?.error}\\n${errorTask?.errorStack}`);\n }\n }\n\n async ai(taskPrompt: string, type = 'action') {\n if (type === 'action') {\n return this.aiAction(taskPrompt);\n }\n if (type === 'query') {\n return this.aiQuery(taskPrompt);\n }\n\n if (type === 'assert') {\n return this.aiAssert(taskPrompt);\n }\n\n if (type === 'tap') {\n return this.aiTap(taskPrompt);\n }\n\n if (type === 'rightClick') {\n return this.aiRightClick(taskPrompt);\n }\n\n throw new Error(\n `Unknown type: ${type}, only support 'action', 'query', 'assert', 'tap', 'rightClick'`,\n );\n }\n\n async runYaml(yamlScriptContent: string): Promise<{\n result: Record<string, any>;\n }> {\n const script = parseYamlScript(yamlScriptContent, 'yaml', true);\n const player = new ScriptPlayer(script, async (target) => {\n return { agent: this, freeFn: [] };\n });\n await player.run();\n\n if (player.status === 'error') {\n const errors = player.taskStatusList\n .filter((task) => task.status === 'error')\n .map((task) => {\n return `task - ${task.name}: ${task.error?.message}`;\n })\n .join('\\n');\n throw new Error(`Error(s) occurred in running yaml script:\\n${errors}`);\n }\n\n return {\n result: player.result,\n };\n }\n\n async evaluateJavaScript(script: string) {\n assert(\n this.page.evaluateJavaScript,\n 'evaluateJavaScript is not supported in current agent',\n );\n return this.page.evaluateJavaScript(script);\n }\n\n async destroy() {\n await this.page.destroy();\n this.resetDump(); // reset dump to release memory\n this.destroyed = true;\n }\n\n async logScreenshot(\n title?: string,\n opt?: {\n content: string;\n },\n ) {\n // 1. screenshot\n const base64 = await this.page.screenshotBase64();\n const now = Date.now();\n // 2. build recorder\n const recorder: ExecutionRecorderItem[] = [\n {\n type: 'screenshot',\n ts: now,\n screenshot: base64,\n },\n ];\n // 3. build ExecutionTaskLog\n const task: ExecutionTaskLog = {\n type: 'Log',\n subType: 'Screenshot',\n status: 'finished',\n recorder,\n timing: {\n start: now,\n end: now,\n cost: 0,\n },\n param: {\n content: opt?.content || '',\n },\n executor: async () => {},\n };\n // 4. build ExecutionDump\n const executionDump: ExecutionDump = {\n sdkVersion: '',\n logTime: now,\n name: `Log - ${title || 'untitled'}`,\n description: opt?.content || '',\n tasks: [task],\n };\n // 5. append to execution dump\n this.appendExecutionDump(executionDump);\n\n try {\n this.onDumpUpdate?.(this.dumpDataString());\n } catch (error) {\n console.error('Failed to update dump', error);\n }\n\n this.writeOutActionDumps();\n }\n\n _unstableLogContent() {\n const { groupName, groupDescription, executions } = this.dump;\n const newExecutions = Array.isArray(executions)\n ? executions.map((execution: any) => {\n const { tasks, ...restExecution } = execution;\n let newTasks = tasks;\n if (Array.isArray(tasks)) {\n newTasks = tasks.map((task: any) => {\n // only remove pageContext and log from task\n const { pageContext, log, ...restTask } = task;\n return restTask;\n });\n }\n return { ...restExecution, ...(newTasks ? { tasks: newTasks } : {}) };\n })\n : [];\n return {\n groupName,\n groupDescription,\n executions: newExecutions,\n };\n }\n\n /**\n * Freezes the current page context to be reused in subsequent AI operations\n * This avoids recalculating page context for each operation\n */\n async freezePageContext(): Promise<void> {\n debug('Freezing page context');\n const context = await this._snapshotContext();\n // Mark the context as frozen\n context._isFrozen = true;\n this.frozenPageContext = context;\n debug('Page context frozen successfully');\n }\n\n /**\n * Unfreezes the page context, allowing AI operations to calculate context dynamically\n */\n async unfreezePageContext(): Promise<void> {\n debug('Unfreezing page context');\n this.frozenPageContext = undefined;\n debug('Page context unfrozen successfully');\n }\n}\n"],"names":["debug","getDebug","distanceOfTwoPoints","p1","p2","x1","y1","x2","y2","Math","includedInRect","point","rect","x","y","left","top","width","height","defaultInsightExtractOption","Agent","action","commonContextParser","prompt","execution","trimmedExecution","trimContextByViewport","currentDump","stringifyDumpData","reportHTMLContent","Error","generateReport","autoPrintReportMsg","writeLogFile","groupedActionDumpFileExt","printReportMsg","task","param","paramStr","tip","typeStr","executor","doNotThrowError","_this","error","console","errorTask","type","opt","actionPlan","plans","Boolean","title","taskTitleStr","locateParamStr","output","locatePrompt","assert","detailedLocateParam","buildDetailedLocateParam","locatePromptOrValue","locatePromptOrOpt","optOrUndefined","value","optWithValue","locatePromptOrKeyName","keyName","undefined","locatePromptOrScrollParam","scrollParam","taskPrompt","_this_taskCache","_this_taskCache1","modelPreferences","cacheable","isVlmUiTars","vlLocateMode","matchedCache","_matchedCache_cacheContent","_matchedCache_cacheContent1","yaml","yamlContent","yamlFlowStr","demand","textPrompt","multimodalPrompt","parsePrompt","center","verifyPrompt","retryLimit","success","retryCount","resultPrompt","deepThink","verifyResult","text","locateOpt","expectCenter","verifyLocateOption","verifyCenter","verifyRect","distance","included","pass","locateParam","locatePlan","locatePlanForLocate","element","assertion","msg","_executor_latestErrorTask","insightOpt","thought","message","yamlScriptContent","script","parseYamlScript","player","ScriptPlayer","target","errors","_task_error","base64","now","Date","recorder","executionDump","groupName","groupDescription","executions","newExecutions","Array","tasks","restExecution","newTasks","pageContext","log","restTask","context","page","opts","Object","globalConfigManager","Insight","TaskCache","MIDSCENE_CACHE","PageTaskExecutor","getReportFileName"],"mappings":";;;;;;;;;;;;;;;;;;;;;AAgEA,MAAMA,QAAQC,SAAS;AAEvB,MAAMC,sBAAsB,CAACC,IAAsBC;IACjD,MAAM,CAACC,IAAIC,GAAG,GAAGH;IACjB,MAAM,CAACI,IAAIC,GAAG,GAAGJ;IACjB,OAAOK,KAAK,KAAK,CAACA,KAAK,IAAI,CAAEJ,AAAAA,CAAAA,KAAKE,EAAC,KAAM,IAAKD,AAAAA,CAAAA,KAAKE,EAAC,KAAM;AAC5D;AAEA,MAAME,iBAAiB,CAACC,OAAyBC;IAC/C,MAAM,CAACC,GAAGC,EAAE,GAAGH;IACf,MAAM,EAAEI,IAAI,EAAEC,GAAG,EAAEC,KAAK,EAAEC,MAAM,EAAE,GAAGN;IACrC,OAAOC,KAAKE,QAAQF,KAAKE,OAAOE,SAASH,KAAKE,OAAOF,KAAKE,MAAME;AAClE;AAEA,MAAMC,8BAAoD;IACxD,aAAa;IACb,oBAAoB;AACtB;AAkBO,MAAMC;IA6EX,MAAM,iBAA0C;QAC9C,OAAO,IAAI,CAAC,IAAI,CAAC,WAAW;IAC9B;IAEA,MAAM,aAAaC,MAAsB,EAAsB;QAE7D,IAAI,IAAI,CAAC,iBAAiB,EAAE;YAC1BrB,MAAM,yCAAyCqB;YAC/C,OAAO,IAAI,CAAC,iBAAiB;QAC/B;QAEA,IAAI,IAAI,CAAC,IAAI,CAAC,UAAU,EAAE;YACxBrB,MAAM,qCAAqCqB;YAC3C,OAAO,MAAM,IAAI,CAAC,IAAI,CAAC,UAAU;QACnC;QACErB,MAAM,yCAAyCqB;QAC/C,OAAO,MAAMC,oBAAoB,IAAI,CAAC,IAAI;IAE9C;IAEA,MAAM,mBAAuC;QAC3C,OAAO,MAAM,IAAI,CAAC,YAAY,CAAC;IACjC;IAEA,MAAM,mBAAmBC,MAAc,EAAE;QACvC,IAAI,CAAC,IAAI,CAAC,eAAe,GAAGA;IAC9B;IAEA,YAAY;QACV,IAAI,CAAC,IAAI,GAAG;YACV,WAAW,IAAI,CAAC,IAAI,CAAC,SAAS;YAC9B,kBAAkB,IAAI,CAAC,IAAI,CAAC,gBAAgB;YAC5C,YAAY,EAAE;YACd,aAAa,EAAE;QACjB;QAEA,OAAO,IAAI,CAAC,IAAI;IAClB;IAEA,oBAAoBC,SAAwB,EAAE;QAE5C,MAAMC,mBAAmBC,sBAAsBF;QAC/C,MAAMG,cAAc,IAAI,CAAC,IAAI;QAC7BA,YAAY,UAAU,CAAC,IAAI,CAACF;IAC9B;IAEA,iBAAiB;QAEf,IAAI,CAAC,IAAI,CAAC,SAAS,GAAG,IAAI,CAAC,IAAI,CAAC,SAAS;QACzC,IAAI,CAAC,IAAI,CAAC,gBAAgB,GAAG,IAAI,CAAC,IAAI,CAAC,gBAAgB;QACvD,OAAOG,kBAAkB,IAAI,CAAC,IAAI;IACpC;IAEA,mBAAmB;QACjB,OAAOC,kBAAkB,IAAI,CAAC,cAAc;IAC9C;IAEA,sBAAsB;QACpB,IAAI,IAAI,CAAC,SAAS,EAChB,MAAM,IAAIC,MACR;QAGJ,MAAM,EAAEC,cAAc,EAAEC,kBAAkB,EAAE,GAAG,IAAI,CAAC,IAAI;QACxD,IAAI,CAAC,UAAU,GAAGC,aAAa;YAC7B,UAAU,IAAI,CAAC,cAAc;YAC7B,SAASC;YACT,aAAa,IAAI,CAAC,cAAc;YAChC,MAAM;YACNH;QACF;QACA/B,MAAM,uBAAuB,IAAI,CAAC,UAAU;QAC5C,IAAI+B,kBAAkBC,sBAAsB,IAAI,CAAC,UAAU,EACzDG,eAAe,IAAI,CAAC,UAAU;IAElC;IAEA,MAAc,uBAAuBC,IAAmB,EAAE;QACxD,MAAMC,QAAQC,SAASF;QACvB,MAAMG,MAAMF,QAAQ,GAAGG,QAAQJ,MAAM,GAAG,EAAEC,OAAO,GAAGG,QAAQJ;QAE5D,IAAI,IAAI,CAAC,cAAc,EACrB,MAAM,IAAI,CAAC,cAAc,CAACG;IAE9B;IAEA,MAAc,iBAAiBE,QAAkB,EAAEC,kBAAkB,KAAK,EAAE;QAC1E,IAAI,CAAC,mBAAmB,CAACD,SAAS,IAAI;QAEtC,IAAI;gBACIE,oBAAAA;YAAN,eAAMA,CAAAA,qBAAAA,AAAAA,CAAAA,QAAAA,IAAI,AAAD,EAAE,YAAY,AAAD,IAAhBA,KAAAA,IAAAA,mBAAAA,IAAAA,CAAAA,OAAoB,IAAI,CAAC,cAAc,GAAE;QACjD,EAAE,OAAOC,OAAO;YACdC,QAAQ,KAAK,CAAC,yBAAyBD;QACzC;QAEA,IAAI,CAAC,mBAAmB;QAExB,IAAIH,SAAS,cAAc,MAAM,CAACC,iBAAiB;YACjD,MAAMI,YAAYL,SAAS,eAAe;YAC1C,MAAM,IAAIX,MAAM,GAAGgB,QAAAA,YAAAA,KAAAA,IAAAA,UAAW,YAAY,CAAC,EAAE,EAAEA,QAAAA,YAAAA,KAAAA,IAAAA,UAAW,UAAU,EAAE,EAAE;gBACtE,OAAOA,QAAAA,YAAAA,KAAAA,IAAAA,UAAW,KAAK;YACzB;QACF;IACF;IAEA,MAAM,wBACJC,IAAY,EACZC,GAAO,EACP;QACAhD,MAAM,2BAA2B+C,MAAM,KAAKC,KAAK,KAAKA;QAEtD,MAAMC,aAAgC;YACpC,MAAMF;YACN,OAAQC,OAAe,CAAC;YACxB,SAAS;QACX;QACAhD,MAAM,cAAciD;QAEpB,MAAMC,QAA0B;YAACD;SAAW,CAAC,MAAM,CACjDE;QAGF,MAAMC,QAAQC,aACZN,MACAO,eAAe,AAACN,CAAAA,QAAAA,MAAAA,KAAAA,IAAAA,IAAa,MAAM,AAAD,KAAK,CAAC;QAG1C,MAAM,EAAEO,MAAM,EAAEd,QAAQ,EAAE,GAAG,MAAM,IAAI,CAAC,YAAY,CAAC,QAAQ,CAACW,OAAOF;QACrE,MAAM,IAAI,CAAC,gBAAgB,CAACT;QAC5B,OAAOc;IACT;IAEA,MAAM,MAAMC,YAAyB,EAAER,GAAkB,EAAE;QACzDS,OAAOD,cAAc;QAErB,MAAME,sBAAsBC,yBAAyBH,cAAcR;QAEnE,OAAO,IAAI,CAAC,uBAAuB,CAAC,OAAO;YACzC,QAAQU;QACV;IACF;IAEA,MAAM,aAAaF,YAAyB,EAAER,GAAkB,EAAE;QAChES,OAAOD,cAAc;QAErB,MAAME,sBAAsBC,yBAAyBH,cAAcR;QAEnE,OAAO,IAAI,CAAC,uBAAuB,CAAC,cAAc;YAChD,QAAQU;QACV;IACF;IAEA,MAAM,QAAQF,YAAyB,EAAER,GAAkB,EAAE;QAC3DS,OAAOD,cAAc;QAErB,MAAME,sBAAsBC,yBAAyBH,cAAcR;QAEnE,OAAO,IAAI,CAAC,uBAAuB,CAAC,SAAS;YAC3C,QAAQU;QACV;IACF;IAmBA,MAAM,QACJE,mBAAyC,EACzCC,iBAGa,EACbC,cAA6B,EAC7B;QACA,IAAIC;QACJ,IAAIP;QACJ,IAAIR;QAKJ,IACE,AAA6B,YAA7B,OAAOa,qBACPA,AAAsB,SAAtBA,qBACA,WAAWA,mBACX;YAEAL,eAAeI;YACf,MAAMI,eAAeH;YAIrBE,QAAQC,aAAa,KAAK;YAC1BhB,MAAMgB;QACR,OAAO;YAELD,QAAQH;YACRJ,eAAeK;YACfb,MAAM;gBACJ,GAAGc,cAAc;gBACjBC;YACF;QACF;QAEAN,OACE,AAAiB,YAAjB,OAAOM,OACP;QAEFN,OAAOD,cAAc;QAErB,MAAME,sBAAsBC,yBAAyBH,cAAcR;QAEnE,OAAO,IAAI,CAAC,uBAAuB,CAAC,SAAS;YAC3C,GAAIA,OAAO,CAAC,CAAC;YACb,QAAQU;QACV;IACF;IAmBA,MAAM,gBACJO,qBAA2C,EAC3CJ,iBAGa,EACbC,cAA6B,EAC7B;QACA,IAAII;QACJ,IAAIV;QACJ,IAAIR;QAGJ,IACE,AAA6B,YAA7B,OAAOa,qBACPA,AAAsB,SAAtBA,qBACA,aAAaA,mBACb;YAEAL,eAAeS;YACfjB,MAAMa;QAGR,OAAO;YAELK,UAAUD;YACVT,eAAeK;YACfb,MAAM;gBACJ,GAAIc,kBAAkB,CAAC,CAAC;gBACxBI;YACF;QACF;QAEAT,OAAOT,QAAAA,MAAAA,KAAAA,IAAAA,IAAK,OAAO,EAAE;QAErB,MAAMU,sBAAsBF,eACxBG,yBAAyBH,cAAcR,OACvCmB;QAEJ,OAAO,IAAI,CAAC,uBAAuB,CAAC,iBAAiB;YACnD,GAAInB,OAAO,CAAC,CAAC;YACb,QAAQU;QACV;IACF;IAmBA,MAAM,SACJU,yBAAgE,EAChEP,iBAAyE,EACzEC,cAA6B,EAC7B;QACA,IAAIO;QACJ,IAAIb;QACJ,IAAIR;QAGJ,IACE,AAA6B,YAA7B,OAAOa,qBACN,gBAAeA,qBACd,gBAAgBA,qBAChB,cAAcA,iBAAgB,GAChC;YAEAL,eAAeY;YACfpB,MAAMa;QACR,OAAO;YAELQ,cAAcD;YACdZ,eAAeK;YACfb,MAAM;gBACJ,GAAIc,kBAAkB,CAAC,CAAC;gBACxB,GAAIO,eAAe,CAAC,CAAC;YACvB;QACF;QAEA,MAAMX,sBAAsBC,yBAC1BH,gBAAgB,IAChBR;QAGF,OAAO,IAAI,CAAC,uBAAuB,CAAC,UAAU;YAC5C,GAAIA,OAAO,CAAC,CAAC;YACb,QAAQU;QACV;IACF;IAEA,MAAM,SACJY,UAAkB,EAClBtB,GAEC,EACD;YAQMuB,iBACcC;QARpB,MAAMC,mBAAsC;YAAE,QAAQ;QAAW;QACjE,MAAMC,YAAY1B,QAAAA,MAAAA,KAAAA,IAAAA,IAAK,SAAS;QAEhC,MAAM2B,cAAcC,AAAmC,kBAAnCA,aAAaH;QACjC,MAAMI,eACJF,eAAeD,AAAc,UAAdA,YACXP,SAAAA,QACAI,CAAAA,kBAAAA,IAAI,CAAC,SAAS,AAAD,IAAbA,KAAAA,IAAAA,gBAAgB,cAAc,CAACD;QACrC,IAAIO,gBAAAA,SAAgBL,CAAAA,mBAAAA,IAAI,CAAC,SAAS,AAAD,IAAbA,KAAAA,IAAAA,iBAAgB,iBAAiB,AAAD,GAAG;gBAInDM,4BAMWC;YARb,MAAM,EAAEtC,QAAQ,EAAE,GAAG,MAAM,IAAI,CAAC,YAAY,CAAC,sBAAsB,CACjE6B,YAAAA,QACAQ,CAAAA,6BAAAA,aAAa,YAAY,AAAD,IAAxBA,KAAAA,IAAAA,2BAA2B,YAAY;YAGzC,MAAM,MAAM,IAAI,CAAC,gBAAgB,CAACrC;YAElCzC,MAAM;YACN,MAAMgF,OAAO,QAAAD,CAAAA,8BAAAA,aAAa,YAAY,AAAD,IAAxBA,KAAAA,IAAAA,4BAA2B,YAAY;YACpD,OAAO,IAAI,CAAC,OAAO,CAACC;QACtB;QAEA,MAAM,EAAEzB,MAAM,EAAEd,QAAQ,EAAE,GAAG,MAAOkC,CAAAA,cAChC,IAAI,CAAC,YAAY,CAAC,YAAY,CAACL,cAC/B,IAAI,CAAC,YAAY,CAAC,MAAM,CAACA,YAAY,IAAI,CAAC,IAAI,CAAC,eAAe;QAGlE,IAAI,IAAI,CAAC,SAAS,IAAIf,CAAAA,QAAAA,SAAAA,KAAAA,IAAAA,OAAQ,QAAQ,AAAD,KAAKmB,AAAc,UAAdA,WAAqB;YAC7D,MAAMO,cAAkC;gBACtC,OAAO;oBACL;wBACE,MAAMX;wBACN,MAAMf,OAAO,QAAQ;oBACvB;iBACD;YACH;YACA,MAAM2B,cAAcF,QAAAA,IAAS,CAACC;YAC9B,IAAI,CAAC,SAAS,CAAC,yBAAyB,CACtC;gBACE,MAAM;gBACN,QAAQX;gBACR,cAAcY;YAChB,GACAL;QAEJ;QAEA,MAAM,IAAI,CAAC,gBAAgB,CAACpC;QAC5B,OAAOc;IACT;IAEA,MAAM,QACJ4B,MAA2B,EAC3BnC,MAA4B7B,2BAA2B,EAClC;QACrB,MAAM,EAAEoC,MAAM,EAAEd,QAAQ,EAAE,GACxB,MAAM,IAAI,CAAC,YAAY,CAAC,wBAAwB,CAAC,SAAS0C,QAAQnC;QACpE,MAAM,IAAI,CAAC,gBAAgB,CAACP;QAC5B,OAAOc;IACT;IAEA,MAAM,UACJhC,MAAmB,EACnByB,MAA4B7B,2BAA2B,EACrC;QAClB,MAAM,EAAEiE,UAAU,EAAEC,gBAAgB,EAAE,GAAGC,YAAY/D;QACrD,MAAM,EAAEgC,MAAM,EAAEd,QAAQ,EAAE,GACxB,MAAM,IAAI,CAAC,YAAY,CAAC,wBAAwB,CAC9C,WACA2C,YACApC,KACAqC;QAEJ,MAAM,IAAI,CAAC,gBAAgB,CAAC5C;QAC5B,OAAOc;IACT;IAEA,MAAM,SACJhC,MAAmB,EACnByB,MAA4B7B,2BAA2B,EACtC;QACjB,MAAM,EAAEiE,UAAU,EAAEC,gBAAgB,EAAE,GAAGC,YAAY/D;QACrD,MAAM,EAAEgC,MAAM,EAAEd,QAAQ,EAAE,GACxB,MAAM,IAAI,CAAC,YAAY,CAAC,wBAAwB,CAC9C,UACA2C,YACApC,KACAqC;QAEJ,MAAM,IAAI,CAAC,gBAAgB,CAAC5C;QAC5B,OAAOc;IACT;IAEA,MAAM,SACJhC,MAAmB,EACnByB,MAA4B7B,2BAA2B,EACtC;QACjB,MAAM,EAAEiE,UAAU,EAAEC,gBAAgB,EAAE,GAAGC,YAAY/D;QACrD,MAAM,EAAEgC,MAAM,EAAEd,QAAQ,EAAE,GACxB,MAAM,IAAI,CAAC,YAAY,CAAC,wBAAwB,CAC9C,UACA2C,YACApC,KACAqC;QAEJ,MAAM,IAAI,CAAC,gBAAgB,CAAC5C;QAC5B,OAAOc;IACT;IAEA,MAAM,MACJhC,MAAmB,EACnByB,MAA4B7B,2BAA2B,EACtC;QACjB,OAAO,IAAI,CAAC,QAAQ,CAACI,QAAQyB;IAC/B;IAEA,MAAM,uBACJuC,MAAwB,EACxBvC,GAI0B,EACkB;QAC5C,MAAM,EAAEwC,eAAe,IAAI,EAAEC,aAAa,CAAC,EAAE,GAAGzC,OAAO,CAAC;QAExD,IAAI0C,UAAU;QACd,IAAIC,aAAa;QACjB,IAAIC,eAAe;QACnB,IAAIC,YAAY7C,AAAAA,CAAAA,QAAAA,MAAAA,KAAAA,IAAAA,IAAK,SAAS,AAAD,KAAK;QAClC,IAAI8C;QAEJ,MAAO,CAACJ,WAAWC,aAAaF,WAAY;YAC1C,IAAIE,cAAc,GAChBE,YAAY;YAEd7F,MACE,cACAuF,QACA,gBACAC,cACA,cACAG,YACA,aACAE;YAEF,MAAME,OAAO,MAAM,IAAI,CAAC,OAAO,CAAC,QAAQ,CAACR,QAAQ;gBAAEM;YAAU;YAC7D7F,MAAM,mBAAmB+F;YACzBtC,OAAOsC,KAAK,WAAW,EAAE,CAAC,+BAA+B,EAAER,OAAO,CAAC,CAAC;YACpEK,eAAeG,KAAK,WAAW;YAE/BD,eAAe,MAAM,IAAI,CAAC,aAAa,CACrCF,cACAC,YAAY;gBAAE,WAAW;YAAK,IAAI1B,QAClCoB,QACAvC;YAEF,IAAI8C,aAAa,IAAI,EACnBJ,UAAU;iBAEVC;QAEJ;QAEA,OAAO;YACL,QAAQC;YACRC;YACAC;QACF;IACF;IAEA,MAAM,cACJvE,MAAc,EACdyE,SAAmC,EACnCC,YAA8B,EAC9BC,kBAA2C,EACX;QAChClG,MAAM,iBAAiBuB,QAAQyE,WAAWC,cAAcC;QAExD,MAAM,EAAE,QAAQC,YAAY,EAAE,MAAMC,UAAU,EAAE,GAAG,MAAM,IAAI,CAAC,QAAQ,CACpE7E,QACAyE;QAEF,MAAMK,WAAWnG,oBAAoB+F,cAAcE;QACnD,MAAMG,WAAW5F,eAAeuF,cAAcG;QAC9C,MAAMG,OACJF,YAAaH,CAAAA,CAAAA,QAAAA,qBAAAA,KAAAA,IAAAA,mBAAoB,uBAAuB,AAAD,KAAK,EAAC,KAC7DI;QACF,MAAMR,eAAe;YACnBS;YACA,MAAMH;YACN,QAAQD;YACR,gBAAgBE;QAClB;QACArG,MAAM,2BAA2B8F;QACjC,OAAOA;IACT;IAEA,MAAM,SAASvE,MAAmB,EAAEyB,GAAkB,EAAE;QACtD,MAAMwD,cAAc7C,yBAAyBpC,QAAQyB;QACrDS,OAAO+C,aAAa;QACpB,MAAMC,aAAaC,oBAAoBF;QACvC,MAAMtD,QAAQ;YAACuD;SAAW;QAC1B,MAAM,EAAEhE,QAAQ,EAAEc,MAAM,EAAE,GAAG,MAAM,IAAI,CAAC,YAAY,CAAC,QAAQ,CAC3DF,aAAa,UAAUC,eAAekD,eACtCtD;QAEF,MAAM,IAAI,CAAC,gBAAgB,CAACT;QAE5B,MAAM,EAAEkE,OAAO,EAAE,GAAGpD;QAEpB,OAAO;YACL,MAAMoD,QAAAA,UAAAA,KAAAA,IAAAA,QAAS,IAAI;YACnB,QAAQA,QAAAA,UAAAA,KAAAA,IAAAA,QAAS,MAAM;YACvB,OAAQ,OAAM,IAAI,CAAC,IAAI,CAAC,IAAI,EAAC,EAAG,GAAG;QACrC;IAGF;IAEA,MAAM,SACJC,SAAsB,EACtBC,GAAY,EACZ7D,GAA2C,EAC3C;YAoBiB8D;QAnBjB,MAAMC,aAAmC;YACvC,aAAa/D,AAAAA,CAAAA,QAAAA,MAAAA,KAAAA,IAAAA,IAAK,WAAW,AAAD,KAAK7B,4BAA4B,WAAW;YACxE,oBACE6B,AAAAA,CAAAA,QAAAA,MAAAA,KAAAA,IAAAA,IAAK,kBAAkB,AAAD,KACtB7B,4BAA4B,kBAAkB;YAChD,eAAe6B,AAAAA,CAAAA,QAAAA,MAAAA,KAAAA,IAAAA,IAAK,aAAa,AAAD,KAAK;YACrC,iBAAiBA,QAAAA,MAAAA,KAAAA,IAAAA,IAAK,eAAe;YACrC,iBAAiBA,QAAAA,MAAAA,KAAAA,IAAAA,IAAK,eAAe;QACvC;QAEA,MAAM,EAAEO,MAAM,EAAEd,QAAQ,EAAEuE,OAAO,EAAE,GAAG,MAAM,IAAI,CAAC,YAAY,CAAC,MAAM,CAClEJ,WACAG;QAEF,MAAM,IAAI,CAAC,gBAAgB,CAACtE,UAAU;QAEtC,MAAMwE,UAAU1D,SACZY,SACA,CAAC,kBAAkB,EAAE0C,OAAOD,UAAU,UAAU,EAC9CI,WAAAA,SAAWF,CAAAA,4BAAAA,SAAS,eAAe,EAAC,IAAzBA,KAAAA,IAAAA,0BAA4B,KAAK,AAAD,KAAK,eAChD;QAEN,IAAI9D,QAAAA,MAAAA,KAAAA,IAAAA,IAAK,eAAe,EACtB,OAAO;YACL,MAAMO;YACNyD;YACAC;QACF;QAGF,IAAI,CAAC1D,QACH,MAAM,IAAIzB,MAAMmF;IAEpB;IAEA,MAAM,UAAUL,SAAsB,EAAE5D,GAAqB,EAAE;QAC7D,MAAM,EAAEP,QAAQ,EAAE,GAAG,MAAM,IAAI,CAAC,YAAY,CAAC,OAAO,CAACmE,WAAW;YAC9D,WAAW5D,AAAAA,CAAAA,QAAAA,MAAAA,KAAAA,IAAAA,IAAK,SAAS,AAAD,KAAK;YAC7B,iBAAiBA,AAAAA,CAAAA,QAAAA,MAAAA,KAAAA,IAAAA,IAAK,eAAe,AAAD,KAAK;QAC3C;QACA,MAAM,IAAI,CAAC,gBAAgB,CAACP,UAAU;QAEtC,IAAIA,SAAS,cAAc,IAAI;YAC7B,MAAMK,YAAYL,SAAS,eAAe;YAC1C,MAAM,IAAIX,MAAM,GAAGgB,QAAAA,YAAAA,KAAAA,IAAAA,UAAW,KAAK,CAAC,EAAE,EAAEA,QAAAA,YAAAA,KAAAA,IAAAA,UAAW,UAAU,EAAE;QACjE;IACF;IAEA,MAAM,GAAGwB,UAAkB,EAAEvB,OAAO,QAAQ,EAAE;QAC5C,IAAIA,AAAS,aAATA,MACF,OAAO,IAAI,CAAC,QAAQ,CAACuB;QAEvB,IAAIvB,AAAS,YAATA,MACF,OAAO,IAAI,CAAC,OAAO,CAACuB;QAGtB,IAAIvB,AAAS,aAATA,MACF,OAAO,IAAI,CAAC,QAAQ,CAACuB;QAGvB,IAAIvB,AAAS,UAATA,MACF,OAAO,IAAI,CAAC,KAAK,CAACuB;QAGpB,IAAIvB,AAAS,iBAATA,MACF,OAAO,IAAI,CAAC,YAAY,CAACuB;QAG3B,MAAM,IAAIxC,MACR,CAAC,cAAc,EAAEiB,KAAK,+DAA+D,CAAC;IAE1F;IAEA,MAAM,QAAQmE,iBAAyB,EAEpC;QACD,MAAMC,SAASC,gBAAgBF,mBAAmB,QAAQ;QAC1D,MAAMG,SAAS,IAAIC,aAAaH,QAAQ,OAAOI,SACtC;gBAAE,OAAO,IAAI;gBAAE,QAAQ,EAAE;YAAC;QAEnC,MAAMF,OAAO,GAAG;QAEhB,IAAIA,AAAkB,YAAlBA,OAAO,MAAM,EAAc;YAC7B,MAAMG,SAASH,OAAO,cAAc,CACjC,MAAM,CAAC,CAACjF,OAASA,AAAgB,YAAhBA,KAAK,MAAM,EAC5B,GAAG,CAAC,CAACA;oBAC2BqF;gBAA/B,OAAO,CAAC,OAAO,EAAErF,KAAK,IAAI,CAAC,EAAE,EAAE,QAAAqF,CAAAA,cAAAA,KAAK,KAAK,AAAD,IAATA,KAAAA,IAAAA,YAAY,OAAO,EAAE;YACtD,GACC,IAAI,CAAC;YACR,MAAM,IAAI3F,MAAM,CAAC,2CAA2C,EAAE0F,QAAQ;QACxE;QAEA,OAAO;YACL,QAAQH,OAAO,MAAM;QACvB;IACF;IAEA,MAAM,mBAAmBF,MAAc,EAAE;QACvC1D,OACE,IAAI,CAAC,IAAI,CAAC,kBAAkB,EAC5B;QAEF,OAAO,IAAI,CAAC,IAAI,CAAC,kBAAkB,CAAC0D;IACtC;IAEA,MAAM,UAAU;QACd,MAAM,IAAI,CAAC,IAAI,CAAC,OAAO;QACvB,IAAI,CAAC,SAAS;QACd,IAAI,CAAC,SAAS,GAAG;IACnB;IAEA,MAAM,cACJ/D,KAAc,EACdJ,GAEC,EACD;QAEA,MAAM0E,SAAS,MAAM,IAAI,CAAC,IAAI,CAAC,gBAAgB;QAC/C,MAAMC,MAAMC,KAAK,GAAG;QAEpB,MAAMC,WAAoC;YACxC;gBACE,MAAM;gBACN,IAAIF;gBACJ,YAAYD;YACd;SACD;QAED,MAAMtF,OAAyB;YAC7B,MAAM;YACN,SAAS;YACT,QAAQ;YACRyF;YACA,QAAQ;gBACN,OAAOF;gBACP,KAAKA;gBACL,MAAM;YACR;YACA,OAAO;gBACL,SAAS3E,AAAAA,CAAAA,QAAAA,MAAAA,KAAAA,IAAAA,IAAK,OAAO,AAAD,KAAK;YAC3B;YACA,UAAU,WAAa;QACzB;QAEA,MAAM8E,gBAA+B;YACnC,YAAY;YACZ,SAASH;YACT,MAAM,CAAC,MAAM,EAAEvE,SAAS,YAAY;YACpC,aAAaJ,AAAAA,CAAAA,QAAAA,MAAAA,KAAAA,IAAAA,IAAK,OAAO,AAAD,KAAK;YAC7B,OAAO;gBAACZ;aAAK;QACf;QAEA,IAAI,CAAC,mBAAmB,CAAC0F;QAEzB,IAAI;gBACFnF,oBAAAA;oBAAAA,CAAAA,qBAAAA,AAAAA,CAAAA,QAAAA,IAAI,AAAD,EAAE,YAAY,AAAD,KAAhBA,mBAAAA,IAAAA,CAAAA,OAAoB,IAAI,CAAC,cAAc;QACzC,EAAE,OAAOC,OAAO;YACdC,QAAQ,KAAK,CAAC,yBAAyBD;QACzC;QAEA,IAAI,CAAC,mBAAmB;IAC1B;IAEA,sBAAsB;QACpB,MAAM,EAAEmF,SAAS,EAAEC,gBAAgB,EAAEC,UAAU,EAAE,GAAG,IAAI,CAAC,IAAI;QAC7D,MAAMC,gBAAgBC,MAAM,OAAO,CAACF,cAChCA,WAAW,GAAG,CAAC,CAACzG;YACd,MAAM,EAAE4G,KAAK,EAAE,GAAGC,eAAe,GAAG7G;YACpC,IAAI8G,WAAWF;YACf,IAAID,MAAM,OAAO,CAACC,QAChBE,WAAWF,MAAM,GAAG,CAAC,CAAChG;gBAEpB,MAAM,EAAEmG,WAAW,EAAEC,GAAG,EAAE,GAAGC,UAAU,GAAGrG;gBAC1C,OAAOqG;YACT;YAEF,OAAO;gBAAE,GAAGJ,aAAa;gBAAE,GAAIC,WAAW;oBAAE,OAAOA;gBAAS,IAAI,CAAC,CAAC;YAAE;QACtE,KACA,EAAE;QACN,OAAO;YACLP;YACAC;YACA,YAAYE;QACd;IACF;IAMA,MAAM,oBAAmC;QACvClI,MAAM;QACN,MAAM0I,UAAU,MAAM,IAAI,CAAC,gBAAgB;QAE3CA,QAAQ,SAAS,GAAG;QACpB,IAAI,CAAC,iBAAiB,GAAGA;QACzB1I,MAAM;IACR;IAKA,MAAM,sBAAqC;QACzCA,MAAM;QACN,IAAI,CAAC,iBAAiB,GAAGmE;QACzBnE,MAAM;IACR;IAh0BA,YAAY2I,IAAc,EAAEC,IAAmB,CAAE;QAhCjD;QAEA;QAEA;QAEA;QAEA;QAEA;QAEA;QAKA,kCAAU;QAEV;QAEA;QAEA;QAEA,oCAAY;QAKZ,uBAAQ,qBAAR;QAGE,IAAI,CAAC,IAAI,GAAGD;QACZ,IAAI,CAAC,IAAI,GAAGE,OAAO,MAAM,CACvB;YACE,gBAAgB;YAChB,oBAAoB;YACpB,WAAW;YACX,kBAAkB;QACpB,GACAD,QAAQ,CAAC;QAEX,IAAIA,AAAAA,CAAAA,QAAAA,OAAAA,KAAAA,IAAAA,KAAM,WAAW,AAAD,KAAK,AAA6B,cAA7B,OAAOA,CAAAA,QAAAA,OAAAA,KAAAA,IAAAA,KAAM,WAAW,AAAD,GAC9C,MAAM,IAAI9G,MACR,CAAC,+DAA+D,EAAE,OAAO8G,CAAAA,QAAAA,OAAAA,KAAAA,IAAAA,KAAM,WAAW,AAAD,GAAG;QAIhGE,oBAAoB,IAAI,CAACF,QAAAA,OAAAA,KAAAA,IAAAA,KAAM,WAAW;QAE1C,IAAI,CAAC,cAAc,GAAG,IAAI,CAAC,IAAI,CAAC,cAAc;QAI9C,IAAI,CAAC,OAAO,GAAG,IAAIG,QAAQ,OAAO1H,SACzB,IAAI,CAAC,YAAY,CAACA;QAG3B,IAAIuH,AAAAA,CAAAA,QAAAA,OAAAA,KAAAA,IAAAA,KAAM,OAAO,AAAD,KAAK,AAAuB,cAAvB,IAAI,CAAC,IAAI,CAAC,QAAQ,EACrC,IAAI,CAAC,SAAS,GAAG,IAAII,UACnBJ,KAAK,OAAO,EACZE,oBAAoB,qBAAqB,CAACG;QAI9C,IAAI,CAAC,YAAY,GAAG,IAAIC,iBAAiB,IAAI,CAAC,IAAI,EAAE,IAAI,CAAC,OAAO,EAAE;YAChE,WAAW,IAAI,CAAC,SAAS;YACzB,aAAa,IAAI,CAAC,sBAAsB,CAAC,IAAI,CAAC,IAAI;QACpD;QACA,IAAI,CAAC,IAAI,GAAG,IAAI,CAAC,SAAS;QAC1B,IAAI,CAAC,cAAc,GACjBN,AAAAA,CAAAA,QAAAA,OAAAA,KAAAA,IAAAA,KAAM,cAAc,AAAD,KACnBO,kBAAkBP,AAAAA,CAAAA,QAAAA,OAAAA,KAAAA,IAAAA,KAAM,MAAM,AAAD,KAAK,IAAI,CAAC,IAAI,CAAC,QAAQ,IAAI;IAC5D;AAuxBF"}
1
+ {"version":3,"file":"agent/agent.mjs","sources":["webpack://@midscene/core/./src/agent/agent.ts"],"sourcesContent":["import {\n type AgentAssertOpt,\n type AgentDescribeElementAtPointResult,\n type AgentWaitForOpt,\n type DeviceAction,\n type ExecutionDump,\n type ExecutionRecorderItem,\n type ExecutionTask,\n type ExecutionTaskLog,\n type Executor,\n type GroupedActionDump,\n Insight,\n type InsightAction,\n type InsightExtractOption,\n type InsightExtractParam,\n type LocateOption,\n type LocateResultElement,\n type LocateValidatorResult,\n type LocatorValidatorOption,\n type MidsceneYamlScript,\n type OnTaskStartTip,\n type PlanningAction,\n type Rect,\n type ScrollParam,\n type TUserPrompt,\n type UIContext,\n} from '../index';\n\nimport yaml from 'js-yaml';\n\nimport {\n groupedActionDumpFileExt,\n reportHTMLContent,\n stringifyDumpData,\n writeLogFile,\n} from '@/utils';\nimport {\n ScriptPlayer,\n buildDetailedLocateParam,\n parseYamlScript,\n} from '../yaml/index';\n\nimport type { AbstractInterface } from '@/device';\nimport {\n type IModelPreferences,\n MIDSCENE_CACHE,\n type TModelConfigFn,\n globalConfigManager,\n vlLocateMode,\n} from '@midscene/shared/env';\nimport { getDebug } from '@midscene/shared/logger';\nimport { assert } from '@midscene/shared/utils';\n// import type { AndroidDeviceInputOpt } from '../device';\nimport { TaskCache } from './task-cache';\nimport { TaskExecutor, locatePlanForLocate } from './tasks';\nimport { locateParamStr, paramStr, taskTitleStr, typeStr } from './ui-utils';\nimport {\n commonContextParser,\n getReportFileName,\n parsePrompt,\n printReportMsg,\n} from './utils';\nimport { trimContextByViewport } from './utils';\n\nconst debug = getDebug('agent');\n\nconst distanceOfTwoPoints = (p1: [number, number], p2: [number, number]) => {\n const [x1, y1] = p1;\n const [x2, y2] = p2;\n return Math.round(Math.sqrt((x1 - x2) ** 2 + (y1 - y2) ** 2));\n};\n\nconst includedInRect = (point: [number, number], rect: Rect) => {\n const [x, y] = point;\n const { left, top, width, height } = rect;\n return x >= left && x <= left + width && y >= top && y <= top + height;\n};\n\nconst defaultInsightExtractOption: InsightExtractOption = {\n domIncluded: false,\n screenshotIncluded: true,\n};\n\nexport interface AgentOpt {\n testId?: string;\n cacheId?: string;\n groupName?: string;\n groupDescription?: string;\n /* if auto generate report, default true */\n generateReport?: boolean;\n /* if auto print report msg, default true */\n autoPrintReportMsg?: boolean;\n onTaskStartTip?: OnTaskStartTip;\n aiActionContext?: string;\n /* custom report file name */\n reportFileName?: string;\n modelConfig?: TModelConfigFn;\n}\n\nexport class Agent<\n InterfaceType extends AbstractInterface = AbstractInterface,\n> {\n interface: InterfaceType;\n\n insight: Insight;\n\n dump: GroupedActionDump;\n\n reportFile?: string | null;\n\n reportFileName?: string;\n\n taskExecutor: TaskExecutor;\n\n opts: AgentOpt;\n\n /**\n * If true, the agent will not perform any actions\n */\n dryMode = false;\n\n onTaskStartTip?: OnTaskStartTip;\n\n taskCache?: TaskCache;\n\n onDumpUpdate?: (dump: string) => void;\n\n destroyed = false;\n\n /**\n * Frozen page context for consistent AI operations\n */\n private frozenUIContext?: UIContext;\n\n // @deprecated use .interface instead\n get page() {\n return this.interface;\n }\n\n constructor(interfaceInstance: InterfaceType, opts?: AgentOpt) {\n this.interface = interfaceInstance;\n this.opts = Object.assign(\n {\n generateReport: true,\n autoPrintReportMsg: true,\n groupName: 'Midscene Report',\n groupDescription: '',\n },\n opts || {},\n );\n if (opts?.modelConfig && typeof opts?.modelConfig !== 'function') {\n throw new Error(\n `opts.modelConfig must be one of function or undefined, but got ${typeof opts?.modelConfig}`,\n );\n }\n\n globalConfigManager.init(opts?.modelConfig);\n\n this.onTaskStartTip = this.opts.onTaskStartTip;\n\n this.insight = new Insight(async (action: InsightAction) => {\n return this.getUIContext(action);\n });\n\n if (opts?.cacheId && this.interface.interfaceType !== 'android') {\n this.taskCache = new TaskCache(\n opts.cacheId,\n globalConfigManager.getEnvConfigInBoolean(MIDSCENE_CACHE), // if we should use cache to match the element\n );\n }\n\n this.taskExecutor = new TaskExecutor(this.interface, this.insight, {\n taskCache: this.taskCache,\n onTaskStart: this.callbackOnTaskStartTip.bind(this),\n });\n this.dump = this.resetDump();\n this.reportFileName =\n opts?.reportFileName ||\n getReportFileName(opts?.testId || this.interface.interfaceType || 'web');\n }\n\n async getActionSpace(): Promise<DeviceAction[]> {\n return this.interface.actionSpace();\n }\n\n async getUIContext(action?: InsightAction): Promise<UIContext> {\n // If page context is frozen, return the frozen context for all actions\n if (this.frozenUIContext) {\n debug('Using frozen page context for action:', action);\n return this.frozenUIContext;\n }\n\n if (this.interface.getContext) {\n debug('Using page.getContext for action:', action);\n return await this.interface.getContext();\n } else {\n debug('Using commonContextParser for action:', action);\n return await commonContextParser(this.interface);\n }\n }\n\n async _snapshotContext(): Promise<UIContext> {\n return await this.getUIContext('locate');\n }\n\n async setAIActionContext(prompt: string) {\n this.opts.aiActionContext = prompt;\n }\n\n resetDump() {\n this.dump = {\n groupName: this.opts.groupName!,\n groupDescription: this.opts.groupDescription,\n executions: [],\n modelBriefs: [],\n };\n\n return this.dump;\n }\n\n appendExecutionDump(execution: ExecutionDump) {\n // use trimContextByViewport to process execution\n const trimmedExecution = trimContextByViewport(execution);\n const currentDump = this.dump;\n currentDump.executions.push(trimmedExecution);\n }\n\n dumpDataString() {\n // update dump info\n this.dump.groupName = this.opts.groupName!;\n this.dump.groupDescription = this.opts.groupDescription;\n return stringifyDumpData(this.dump);\n }\n\n reportHTMLString() {\n return reportHTMLContent(this.dumpDataString());\n }\n\n writeOutActionDumps() {\n if (this.destroyed) {\n throw new Error(\n 'PageAgent has been destroyed. Cannot update report file.',\n );\n }\n const { generateReport, autoPrintReportMsg } = this.opts;\n this.reportFile = writeLogFile({\n fileName: this.reportFileName!,\n fileExt: groupedActionDumpFileExt,\n fileContent: this.dumpDataString(),\n type: 'dump',\n generateReport,\n });\n debug('writeOutActionDumps', this.reportFile);\n if (generateReport && autoPrintReportMsg && this.reportFile) {\n printReportMsg(this.reportFile);\n }\n }\n\n private async callbackOnTaskStartTip(task: ExecutionTask) {\n const param = paramStr(task);\n const tip = param ? `${typeStr(task)} - ${param}` : typeStr(task);\n\n if (this.onTaskStartTip) {\n await this.onTaskStartTip(tip);\n }\n }\n\n private async afterTaskRunning(executor: Executor, doNotThrowError = false) {\n this.appendExecutionDump(executor.dump());\n\n try {\n await this.onDumpUpdate?.(this.dumpDataString());\n } catch (error) {\n console.error('Error in onDumpUpdate', error);\n }\n\n this.writeOutActionDumps();\n\n if (executor.isInErrorState() && !doNotThrowError) {\n const errorTask = executor.latestErrorTask();\n throw new Error(`${errorTask?.errorMessage}\\n${errorTask?.errorStack}`, {\n cause: errorTask?.error,\n });\n }\n }\n\n async callActionInActionSpace<T = any>(\n type: string,\n opt?: T, // and all other action params\n ) {\n debug('callActionInActionSpace', type, ',', opt, ',', opt);\n\n const actionPlan: PlanningAction<T> = {\n type: type as any,\n param: (opt as any) || {},\n thought: '',\n };\n debug('actionPlan', actionPlan); // , ', in which the locateParam is', locateParam);\n\n const plans: PlanningAction[] = [actionPlan].filter(\n Boolean,\n ) as PlanningAction[];\n\n const title = taskTitleStr(\n type as any,\n locateParamStr((opt as any)?.locate || {}),\n );\n\n const { output, executor } = await this.taskExecutor.runPlans(title, plans);\n await this.afterTaskRunning(executor);\n return output;\n }\n\n async aiTap(locatePrompt: TUserPrompt, opt?: LocateOption) {\n assert(locatePrompt, 'missing locate prompt for tap');\n\n const detailedLocateParam = buildDetailedLocateParam(locatePrompt, opt);\n\n return this.callActionInActionSpace('Tap', {\n locate: detailedLocateParam,\n });\n }\n\n async aiRightClick(locatePrompt: TUserPrompt, opt?: LocateOption) {\n assert(locatePrompt, 'missing locate prompt for right click');\n\n const detailedLocateParam = buildDetailedLocateParam(locatePrompt, opt);\n\n return this.callActionInActionSpace('RightClick', {\n locate: detailedLocateParam,\n });\n }\n\n async aiHover(locatePrompt: TUserPrompt, opt?: LocateOption) {\n assert(locatePrompt, 'missing locate prompt for hover');\n\n const detailedLocateParam = buildDetailedLocateParam(locatePrompt, opt);\n\n return this.callActionInActionSpace('Hover', {\n locate: detailedLocateParam,\n });\n }\n\n // New signature, always use locatePrompt as the first param\n async aiInput(\n locatePrompt: TUserPrompt,\n opt: LocateOption & { value: string }, // AndroidDeviceInputOpt &\n ): Promise<any>;\n\n // Legacy signature - deprecated\n /**\n * @deprecated Use aiInput(locatePrompt, opt) instead where opt contains the value\n */\n async aiInput(\n value: string,\n locatePrompt: TUserPrompt,\n opt?: LocateOption, // AndroidDeviceInputOpt &\n ): Promise<any>;\n\n // Implementation\n async aiInput(\n locatePromptOrValue: TUserPrompt | string,\n locatePromptOrOpt:\n | TUserPrompt\n | (LocateOption & { value: string }) // AndroidDeviceInputOpt &\n | undefined,\n optOrUndefined?: LocateOption, // AndroidDeviceInputOpt &\n ) {\n let value: string;\n let locatePrompt: TUserPrompt;\n let opt:\n | (LocateOption & { value: string }) // AndroidDeviceInputOpt &\n | undefined;\n\n // Check if using new signature (first param is locatePrompt, second has value)\n if (\n typeof locatePromptOrOpt === 'object' &&\n locatePromptOrOpt !== null &&\n 'value' in locatePromptOrOpt\n ) {\n // New signature: aiInput(locatePrompt, opt)\n locatePrompt = locatePromptOrValue as TUserPrompt;\n const optWithValue = locatePromptOrOpt as LocateOption & {\n // AndroidDeviceInputOpt &\n value: string;\n };\n value = optWithValue.value;\n opt = optWithValue;\n } else {\n // Legacy signature: aiInput(value, locatePrompt, opt)\n value = locatePromptOrValue as string;\n locatePrompt = locatePromptOrOpt as TUserPrompt;\n opt = {\n ...optOrUndefined,\n value,\n };\n }\n\n assert(\n typeof value === 'string',\n 'input value must be a string, use empty string if you want to clear the input',\n );\n assert(locatePrompt, 'missing locate prompt for input');\n\n const detailedLocateParam = buildDetailedLocateParam(locatePrompt, opt);\n\n return this.callActionInActionSpace('Input', {\n ...(opt || {}),\n locate: detailedLocateParam,\n });\n }\n\n // New signature\n async aiKeyboardPress(\n locatePrompt: TUserPrompt,\n opt: LocateOption & { keyName: string },\n ): Promise<any>;\n\n // Legacy signature - deprecated\n /**\n * @deprecated Use aiKeyboardPress(locatePrompt, opt) instead where opt contains the keyName\n */\n async aiKeyboardPress(\n keyName: string,\n locatePrompt?: TUserPrompt,\n opt?: LocateOption,\n ): Promise<any>;\n\n // Implementation\n async aiKeyboardPress(\n locatePromptOrKeyName: TUserPrompt | string,\n locatePromptOrOpt:\n | TUserPrompt\n | (LocateOption & { keyName: string })\n | undefined,\n optOrUndefined?: LocateOption,\n ) {\n let keyName: string;\n let locatePrompt: TUserPrompt | undefined;\n let opt: (LocateOption & { keyName: string }) | undefined;\n\n // Check if using new signature (first param is locatePrompt, second has keyName)\n if (\n typeof locatePromptOrOpt === 'object' &&\n locatePromptOrOpt !== null &&\n 'keyName' in locatePromptOrOpt\n ) {\n // New signature: aiKeyboardPress(locatePrompt, opt)\n locatePrompt = locatePromptOrKeyName as TUserPrompt;\n opt = locatePromptOrOpt as LocateOption & {\n keyName: string;\n };\n } else {\n // Legacy signature: aiKeyboardPress(keyName, locatePrompt, opt)\n keyName = locatePromptOrKeyName as string;\n locatePrompt = locatePromptOrOpt as TUserPrompt | undefined;\n opt = {\n ...(optOrUndefined || {}),\n keyName,\n };\n }\n\n assert(opt?.keyName, 'missing keyName for keyboard press');\n\n const detailedLocateParam = locatePrompt\n ? buildDetailedLocateParam(locatePrompt, opt)\n : undefined;\n\n return this.callActionInActionSpace('KeyboardPress', {\n ...(opt || {}),\n locate: detailedLocateParam,\n });\n }\n\n // New signature\n async aiScroll(\n locatePrompt: TUserPrompt | undefined,\n opt: LocateOption & ScrollParam,\n ): Promise<any>;\n\n // Legacy signature - deprecated\n /**\n * @deprecated Use aiScroll(locatePrompt, opt) instead where opt contains the scroll parameters\n */\n async aiScroll(\n scrollParam: ScrollParam,\n locatePrompt?: TUserPrompt,\n opt?: LocateOption,\n ): Promise<any>;\n\n // Implementation\n async aiScroll(\n locatePromptOrScrollParam: TUserPrompt | ScrollParam | undefined,\n locatePromptOrOpt: TUserPrompt | (LocateOption & ScrollParam) | undefined,\n optOrUndefined?: LocateOption,\n ) {\n let scrollParam: ScrollParam;\n let locatePrompt: TUserPrompt | undefined;\n let opt: LocateOption | undefined;\n\n // Check if using new signature (first param is locatePrompt, second has scroll params)\n if (\n typeof locatePromptOrOpt === 'object' &&\n ('direction' in locatePromptOrOpt ||\n 'scrollType' in locatePromptOrOpt ||\n 'distance' in locatePromptOrOpt)\n ) {\n // New signature: aiScroll(locatePrompt, opt)\n locatePrompt = locatePromptOrScrollParam as TUserPrompt;\n opt = locatePromptOrOpt as LocateOption & ScrollParam;\n } else {\n // Legacy signature: aiScroll(scrollParam, locatePrompt, opt)\n scrollParam = locatePromptOrScrollParam as ScrollParam;\n locatePrompt = locatePromptOrOpt as TUserPrompt | undefined;\n opt = {\n ...(optOrUndefined || {}),\n ...(scrollParam || {}),\n };\n }\n\n const detailedLocateParam = buildDetailedLocateParam(\n locatePrompt || '',\n opt,\n );\n\n return this.callActionInActionSpace('Scroll', {\n ...(opt || {}),\n locate: detailedLocateParam,\n });\n }\n\n async aiAction(\n taskPrompt: string,\n opt?: {\n cacheable?: boolean;\n },\n ) {\n const modelPreferences: IModelPreferences = { intent: 'planning' };\n const cacheable = opt?.cacheable;\n // if vlm-ui-tars, plan cache is not used\n const isVlmUiTars = vlLocateMode(modelPreferences) === 'vlm-ui-tars';\n const matchedCache =\n isVlmUiTars || cacheable === false\n ? undefined\n : this.taskCache?.matchPlanCache(taskPrompt);\n if (matchedCache && this.taskCache?.isCacheResultUsed) {\n // log into report file\n const { executor } = await this.taskExecutor.loadYamlFlowAsPlanning(\n taskPrompt,\n matchedCache.cacheContent?.yamlWorkflow,\n );\n\n await await this.afterTaskRunning(executor);\n\n debug('matched cache, will call .runYaml to run the action');\n const yaml = matchedCache.cacheContent?.yamlWorkflow;\n return this.runYaml(yaml);\n }\n\n const { output, executor } = await (isVlmUiTars\n ? this.taskExecutor.actionToGoal(taskPrompt)\n : this.taskExecutor.action(taskPrompt, this.opts.aiActionContext));\n\n // update cache\n if (this.taskCache && output?.yamlFlow && cacheable !== false) {\n const yamlContent: MidsceneYamlScript = {\n tasks: [\n {\n name: taskPrompt,\n flow: output.yamlFlow,\n },\n ],\n };\n const yamlFlowStr = yaml.dump(yamlContent);\n this.taskCache.updateOrAppendCacheRecord(\n {\n type: 'plan',\n prompt: taskPrompt,\n yamlWorkflow: yamlFlowStr,\n },\n matchedCache,\n );\n }\n\n await this.afterTaskRunning(executor);\n return output;\n }\n\n async aiQuery<ReturnType = any>(\n demand: InsightExtractParam,\n opt: InsightExtractOption = defaultInsightExtractOption,\n ): Promise<ReturnType> {\n const { output, executor } =\n await this.taskExecutor.createTypeQueryExecution('Query', demand, opt);\n await this.afterTaskRunning(executor);\n return output as ReturnType;\n }\n\n async aiBoolean(\n prompt: TUserPrompt,\n opt: InsightExtractOption = defaultInsightExtractOption,\n ): Promise<boolean> {\n const { textPrompt, multimodalPrompt } = parsePrompt(prompt);\n const { output, executor } =\n await this.taskExecutor.createTypeQueryExecution(\n 'Boolean',\n textPrompt,\n opt,\n multimodalPrompt,\n );\n await this.afterTaskRunning(executor);\n return output as boolean;\n }\n\n async aiNumber(\n prompt: TUserPrompt,\n opt: InsightExtractOption = defaultInsightExtractOption,\n ): Promise<number> {\n const { textPrompt, multimodalPrompt } = parsePrompt(prompt);\n const { output, executor } =\n await this.taskExecutor.createTypeQueryExecution(\n 'Number',\n textPrompt,\n opt,\n multimodalPrompt,\n );\n await this.afterTaskRunning(executor);\n return output as number;\n }\n\n async aiString(\n prompt: TUserPrompt,\n opt: InsightExtractOption = defaultInsightExtractOption,\n ): Promise<string> {\n const { textPrompt, multimodalPrompt } = parsePrompt(prompt);\n const { output, executor } =\n await this.taskExecutor.createTypeQueryExecution(\n 'String',\n textPrompt,\n opt,\n multimodalPrompt,\n );\n await this.afterTaskRunning(executor);\n return output as string;\n }\n\n async aiAsk(\n prompt: TUserPrompt,\n opt: InsightExtractOption = defaultInsightExtractOption,\n ): Promise<string> {\n return this.aiString(prompt, opt);\n }\n\n async describeElementAtPoint(\n center: [number, number],\n opt?: {\n verifyPrompt?: boolean;\n retryLimit?: number;\n deepThink?: boolean;\n } & LocatorValidatorOption,\n ): Promise<AgentDescribeElementAtPointResult> {\n const { verifyPrompt = true, retryLimit = 3 } = opt || {};\n\n let success = false;\n let retryCount = 0;\n let resultPrompt = '';\n let deepThink = opt?.deepThink || false;\n let verifyResult: LocateValidatorResult | undefined;\n\n while (!success && retryCount < retryLimit) {\n if (retryCount >= 2) {\n deepThink = true;\n }\n debug(\n 'aiDescribe',\n center,\n 'verifyPrompt',\n verifyPrompt,\n 'retryCount',\n retryCount,\n 'deepThink',\n deepThink,\n );\n const text = await this.insight.describe(center, { deepThink });\n debug('aiDescribe text', text);\n assert(text.description, `failed to describe element at [${center}]`);\n resultPrompt = text.description;\n\n verifyResult = await this.verifyLocator(\n resultPrompt,\n deepThink ? { deepThink: true } : undefined,\n center,\n opt,\n );\n if (verifyResult.pass) {\n success = true;\n } else {\n retryCount++;\n }\n }\n\n return {\n prompt: resultPrompt,\n deepThink,\n verifyResult,\n };\n }\n\n async verifyLocator(\n prompt: string,\n locateOpt: LocateOption | undefined,\n expectCenter: [number, number],\n verifyLocateOption?: LocatorValidatorOption,\n ): Promise<LocateValidatorResult> {\n debug('verifyLocator', prompt, locateOpt, expectCenter, verifyLocateOption);\n\n const { center: verifyCenter, rect: verifyRect } = await this.aiLocate(\n prompt,\n locateOpt,\n );\n const distance = distanceOfTwoPoints(expectCenter, verifyCenter);\n const included = includedInRect(expectCenter, verifyRect);\n const pass =\n distance <= (verifyLocateOption?.centerDistanceThreshold || 20) ||\n included;\n const verifyResult = {\n pass,\n rect: verifyRect,\n center: verifyCenter,\n centerDistance: distance,\n };\n debug('aiDescribe verifyResult', verifyResult);\n return verifyResult;\n }\n\n async aiLocate(prompt: TUserPrompt, opt?: LocateOption) {\n const locateParam = buildDetailedLocateParam(prompt, opt);\n assert(locateParam, 'cannot get locate param for aiLocate');\n const locatePlan = locatePlanForLocate(locateParam);\n const plans = [locatePlan];\n const { executor, output } = await this.taskExecutor.runPlans(\n taskTitleStr('Locate', locateParamStr(locateParam)),\n plans,\n );\n await this.afterTaskRunning(executor);\n\n const { element } = output;\n\n return {\n rect: element?.rect,\n center: element?.center,\n scale: (await this.interface.size()).dpr,\n } as Pick<LocateResultElement, 'rect' | 'center'> & {\n scale: number;\n };\n }\n\n async aiAssert(\n assertion: TUserPrompt,\n msg?: string,\n opt?: AgentAssertOpt & InsightExtractOption,\n ) {\n const insightOpt: InsightExtractOption = {\n domIncluded: opt?.domIncluded ?? defaultInsightExtractOption.domIncluded,\n screenshotIncluded:\n opt?.screenshotIncluded ??\n defaultInsightExtractOption.screenshotIncluded,\n returnThought: opt?.returnThought ?? true,\n isWaitForAssert: opt?.isWaitForAssert,\n doNotThrowError: opt?.doNotThrowError,\n };\n\n const { output, executor, thought } = await this.taskExecutor.assert(\n assertion,\n insightOpt,\n );\n await this.afterTaskRunning(executor, true);\n\n const message = output\n ? undefined\n : `Assertion failed: ${msg || assertion}\\nReason: ${\n thought || executor.latestErrorTask()?.error || '(no_reason)'\n }`;\n\n if (opt?.keepRawResponse) {\n return {\n pass: output,\n thought,\n message,\n };\n }\n\n if (!output) {\n throw new Error(message);\n }\n }\n\n async aiWaitFor(assertion: TUserPrompt, opt?: AgentWaitForOpt) {\n const { executor } = await this.taskExecutor.waitFor(assertion, {\n timeoutMs: opt?.timeoutMs || 15 * 1000,\n checkIntervalMs: opt?.checkIntervalMs || 3 * 1000,\n });\n await this.afterTaskRunning(executor, true);\n\n if (executor.isInErrorState()) {\n const errorTask = executor.latestErrorTask();\n throw new Error(`${errorTask?.error}\\n${errorTask?.errorStack}`);\n }\n }\n\n async ai(taskPrompt: string, type = 'action') {\n if (type === 'action') {\n return this.aiAction(taskPrompt);\n }\n if (type === 'query') {\n return this.aiQuery(taskPrompt);\n }\n\n if (type === 'assert') {\n return this.aiAssert(taskPrompt);\n }\n\n if (type === 'tap') {\n return this.aiTap(taskPrompt);\n }\n\n if (type === 'rightClick') {\n return this.aiRightClick(taskPrompt);\n }\n\n throw new Error(\n `Unknown type: ${type}, only support 'action', 'query', 'assert', 'tap', 'rightClick'`,\n );\n }\n\n async runYaml(yamlScriptContent: string): Promise<{\n result: Record<string, any>;\n }> {\n const script = parseYamlScript(yamlScriptContent, 'yaml', true);\n const player = new ScriptPlayer(script, async (target) => {\n return { agent: this, freeFn: [] };\n });\n await player.run();\n\n if (player.status === 'error') {\n const errors = player.taskStatusList\n .filter((task) => task.status === 'error')\n .map((task) => {\n return `task - ${task.name}: ${task.error?.message}`;\n })\n .join('\\n');\n throw new Error(`Error(s) occurred in running yaml script:\\n${errors}`);\n }\n\n return {\n result: player.result,\n };\n }\n\n async evaluateJavaScript(script: string) {\n assert(\n this.interface.evaluateJavaScript,\n 'evaluateJavaScript is not supported in current agent',\n );\n return this.interface.evaluateJavaScript(script);\n }\n\n async destroy() {\n await this.interface.destroy();\n this.resetDump(); // reset dump to release memory\n this.destroyed = true;\n }\n\n async logScreenshot(\n title?: string,\n opt?: {\n content: string;\n },\n ) {\n // 1. screenshot\n const base64 = await this.interface.screenshotBase64();\n const now = Date.now();\n // 2. build recorder\n const recorder: ExecutionRecorderItem[] = [\n {\n type: 'screenshot',\n ts: now,\n screenshot: base64,\n },\n ];\n // 3. build ExecutionTaskLog\n const task: ExecutionTaskLog = {\n type: 'Log',\n subType: 'Screenshot',\n status: 'finished',\n recorder,\n timing: {\n start: now,\n end: now,\n cost: 0,\n },\n param: {\n content: opt?.content || '',\n },\n executor: async () => {},\n };\n // 4. build ExecutionDump\n const executionDump: ExecutionDump = {\n sdkVersion: '',\n logTime: now,\n name: `Log - ${title || 'untitled'}`,\n description: opt?.content || '',\n tasks: [task],\n };\n // 5. append to execution dump\n this.appendExecutionDump(executionDump);\n\n try {\n this.onDumpUpdate?.(this.dumpDataString());\n } catch (error) {\n console.error('Failed to update dump', error);\n }\n\n this.writeOutActionDumps();\n }\n\n _unstableLogContent() {\n const { groupName, groupDescription, executions } = this.dump;\n const newExecutions = Array.isArray(executions)\n ? executions.map((execution: any) => {\n const { tasks, ...restExecution } = execution;\n let newTasks = tasks;\n if (Array.isArray(tasks)) {\n newTasks = tasks.map((task: any) => {\n // only remove uiContext and log from task\n const { uiContext, log, ...restTask } = task;\n return restTask;\n });\n }\n return { ...restExecution, ...(newTasks ? { tasks: newTasks } : {}) };\n })\n : [];\n return {\n groupName,\n groupDescription,\n executions: newExecutions,\n };\n }\n\n /**\n * Freezes the current page context to be reused in subsequent AI operations\n * This avoids recalculating page context for each operation\n */\n async freezePageContext(): Promise<void> {\n debug('Freezing page context');\n const context = await this._snapshotContext();\n // Mark the context as frozen\n context._isFrozen = true;\n this.frozenUIContext = context;\n debug('Page context frozen successfully');\n }\n\n /**\n * Unfreezes the page context, allowing AI operations to calculate context dynamically\n */\n async unfreezePageContext(): Promise<void> {\n debug('Unfreezing page context');\n this.frozenUIContext = undefined;\n debug('Page context unfrozen successfully');\n }\n}\n"],"names":["debug","getDebug","distanceOfTwoPoints","p1","p2","x1","y1","x2","y2","Math","includedInRect","point","rect","x","y","left","top","width","height","defaultInsightExtractOption","Agent","action","commonContextParser","prompt","execution","trimmedExecution","trimContextByViewport","currentDump","stringifyDumpData","reportHTMLContent","Error","generateReport","autoPrintReportMsg","writeLogFile","groupedActionDumpFileExt","printReportMsg","task","param","paramStr","tip","typeStr","executor","doNotThrowError","_this","error","console","errorTask","type","opt","actionPlan","plans","Boolean","title","taskTitleStr","locateParamStr","output","locatePrompt","assert","detailedLocateParam","buildDetailedLocateParam","locatePromptOrValue","locatePromptOrOpt","optOrUndefined","value","optWithValue","locatePromptOrKeyName","keyName","undefined","locatePromptOrScrollParam","scrollParam","taskPrompt","_this_taskCache","_this_taskCache1","modelPreferences","cacheable","isVlmUiTars","vlLocateMode","matchedCache","_matchedCache_cacheContent","_matchedCache_cacheContent1","yaml","yamlContent","yamlFlowStr","demand","textPrompt","multimodalPrompt","parsePrompt","center","verifyPrompt","retryLimit","success","retryCount","resultPrompt","deepThink","verifyResult","text","locateOpt","expectCenter","verifyLocateOption","verifyCenter","verifyRect","distance","included","pass","locateParam","locatePlan","locatePlanForLocate","element","assertion","msg","_executor_latestErrorTask","insightOpt","thought","message","yamlScriptContent","script","parseYamlScript","player","ScriptPlayer","target","errors","_task_error","base64","now","Date","recorder","executionDump","groupName","groupDescription","executions","newExecutions","Array","tasks","restExecution","newTasks","uiContext","log","restTask","context","interfaceInstance","opts","Object","globalConfigManager","Insight","TaskCache","MIDSCENE_CACHE","TaskExecutor","getReportFileName"],"mappings":";;;;;;;;;;;;;;;;;;;;;AAgEA,MAAMA,QAAQC,SAAS;AAEvB,MAAMC,sBAAsB,CAACC,IAAsBC;IACjD,MAAM,CAACC,IAAIC,GAAG,GAAGH;IACjB,MAAM,CAACI,IAAIC,GAAG,GAAGJ;IACjB,OAAOK,KAAK,KAAK,CAACA,KAAK,IAAI,CAAEJ,AAAAA,CAAAA,KAAKE,EAAC,KAAM,IAAKD,AAAAA,CAAAA,KAAKE,EAAC,KAAM;AAC5D;AAEA,MAAME,iBAAiB,CAACC,OAAyBC;IAC/C,MAAM,CAACC,GAAGC,EAAE,GAAGH;IACf,MAAM,EAAEI,IAAI,EAAEC,GAAG,EAAEC,KAAK,EAAEC,MAAM,EAAE,GAAGN;IACrC,OAAOC,KAAKE,QAAQF,KAAKE,OAAOE,SAASH,KAAKE,OAAOF,KAAKE,MAAME;AAClE;AAEA,MAAMC,8BAAoD;IACxD,aAAa;IACb,oBAAoB;AACtB;AAkBO,MAAMC;IAoCX,IAAI,OAAO;QACT,OAAO,IAAI,CAAC,SAAS;IACvB;IA4CA,MAAM,iBAA0C;QAC9C,OAAO,IAAI,CAAC,SAAS,CAAC,WAAW;IACnC;IAEA,MAAM,aAAaC,MAAsB,EAAsB;QAE7D,IAAI,IAAI,CAAC,eAAe,EAAE;YACxBrB,MAAM,yCAAyCqB;YAC/C,OAAO,IAAI,CAAC,eAAe;QAC7B;QAEA,IAAI,IAAI,CAAC,SAAS,CAAC,UAAU,EAAE;YAC7BrB,MAAM,qCAAqCqB;YAC3C,OAAO,MAAM,IAAI,CAAC,SAAS,CAAC,UAAU;QACxC;QACErB,MAAM,yCAAyCqB;QAC/C,OAAO,MAAMC,oBAAoB,IAAI,CAAC,SAAS;IAEnD;IAEA,MAAM,mBAAuC;QAC3C,OAAO,MAAM,IAAI,CAAC,YAAY,CAAC;IACjC;IAEA,MAAM,mBAAmBC,MAAc,EAAE;QACvC,IAAI,CAAC,IAAI,CAAC,eAAe,GAAGA;IAC9B;IAEA,YAAY;QACV,IAAI,CAAC,IAAI,GAAG;YACV,WAAW,IAAI,CAAC,IAAI,CAAC,SAAS;YAC9B,kBAAkB,IAAI,CAAC,IAAI,CAAC,gBAAgB;YAC5C,YAAY,EAAE;YACd,aAAa,EAAE;QACjB;QAEA,OAAO,IAAI,CAAC,IAAI;IAClB;IAEA,oBAAoBC,SAAwB,EAAE;QAE5C,MAAMC,mBAAmBC,sBAAsBF;QAC/C,MAAMG,cAAc,IAAI,CAAC,IAAI;QAC7BA,YAAY,UAAU,CAAC,IAAI,CAACF;IAC9B;IAEA,iBAAiB;QAEf,IAAI,CAAC,IAAI,CAAC,SAAS,GAAG,IAAI,CAAC,IAAI,CAAC,SAAS;QACzC,IAAI,CAAC,IAAI,CAAC,gBAAgB,GAAG,IAAI,CAAC,IAAI,CAAC,gBAAgB;QACvD,OAAOG,kBAAkB,IAAI,CAAC,IAAI;IACpC;IAEA,mBAAmB;QACjB,OAAOC,kBAAkB,IAAI,CAAC,cAAc;IAC9C;IAEA,sBAAsB;QACpB,IAAI,IAAI,CAAC,SAAS,EAChB,MAAM,IAAIC,MACR;QAGJ,MAAM,EAAEC,cAAc,EAAEC,kBAAkB,EAAE,GAAG,IAAI,CAAC,IAAI;QACxD,IAAI,CAAC,UAAU,GAAGC,aAAa;YAC7B,UAAU,IAAI,CAAC,cAAc;YAC7B,SAASC;YACT,aAAa,IAAI,CAAC,cAAc;YAChC,MAAM;YACNH;QACF;QACA/B,MAAM,uBAAuB,IAAI,CAAC,UAAU;QAC5C,IAAI+B,kBAAkBC,sBAAsB,IAAI,CAAC,UAAU,EACzDG,eAAe,IAAI,CAAC,UAAU;IAElC;IAEA,MAAc,uBAAuBC,IAAmB,EAAE;QACxD,MAAMC,QAAQC,SAASF;QACvB,MAAMG,MAAMF,QAAQ,GAAGG,QAAQJ,MAAM,GAAG,EAAEC,OAAO,GAAGG,QAAQJ;QAE5D,IAAI,IAAI,CAAC,cAAc,EACrB,MAAM,IAAI,CAAC,cAAc,CAACG;IAE9B;IAEA,MAAc,iBAAiBE,QAAkB,EAAEC,kBAAkB,KAAK,EAAE;QAC1E,IAAI,CAAC,mBAAmB,CAACD,SAAS,IAAI;QAEtC,IAAI;gBACIE,oBAAAA;YAAN,eAAMA,CAAAA,qBAAAA,AAAAA,CAAAA,QAAAA,IAAI,AAAD,EAAE,YAAY,AAAD,IAAhBA,KAAAA,IAAAA,mBAAAA,IAAAA,CAAAA,OAAoB,IAAI,CAAC,cAAc,GAAE;QACjD,EAAE,OAAOC,OAAO;YACdC,QAAQ,KAAK,CAAC,yBAAyBD;QACzC;QAEA,IAAI,CAAC,mBAAmB;QAExB,IAAIH,SAAS,cAAc,MAAM,CAACC,iBAAiB;YACjD,MAAMI,YAAYL,SAAS,eAAe;YAC1C,MAAM,IAAIX,MAAM,GAAGgB,QAAAA,YAAAA,KAAAA,IAAAA,UAAW,YAAY,CAAC,EAAE,EAAEA,QAAAA,YAAAA,KAAAA,IAAAA,UAAW,UAAU,EAAE,EAAE;gBACtE,OAAOA,QAAAA,YAAAA,KAAAA,IAAAA,UAAW,KAAK;YACzB;QACF;IACF;IAEA,MAAM,wBACJC,IAAY,EACZC,GAAO,EACP;QACAhD,MAAM,2BAA2B+C,MAAM,KAAKC,KAAK,KAAKA;QAEtD,MAAMC,aAAgC;YACpC,MAAMF;YACN,OAAQC,OAAe,CAAC;YACxB,SAAS;QACX;QACAhD,MAAM,cAAciD;QAEpB,MAAMC,QAA0B;YAACD;SAAW,CAAC,MAAM,CACjDE;QAGF,MAAMC,QAAQC,aACZN,MACAO,eAAe,AAACN,CAAAA,QAAAA,MAAAA,KAAAA,IAAAA,IAAa,MAAM,AAAD,KAAK,CAAC;QAG1C,MAAM,EAAEO,MAAM,EAAEd,QAAQ,EAAE,GAAG,MAAM,IAAI,CAAC,YAAY,CAAC,QAAQ,CAACW,OAAOF;QACrE,MAAM,IAAI,CAAC,gBAAgB,CAACT;QAC5B,OAAOc;IACT;IAEA,MAAM,MAAMC,YAAyB,EAAER,GAAkB,EAAE;QACzDS,OAAOD,cAAc;QAErB,MAAME,sBAAsBC,yBAAyBH,cAAcR;QAEnE,OAAO,IAAI,CAAC,uBAAuB,CAAC,OAAO;YACzC,QAAQU;QACV;IACF;IAEA,MAAM,aAAaF,YAAyB,EAAER,GAAkB,EAAE;QAChES,OAAOD,cAAc;QAErB,MAAME,sBAAsBC,yBAAyBH,cAAcR;QAEnE,OAAO,IAAI,CAAC,uBAAuB,CAAC,cAAc;YAChD,QAAQU;QACV;IACF;IAEA,MAAM,QAAQF,YAAyB,EAAER,GAAkB,EAAE;QAC3DS,OAAOD,cAAc;QAErB,MAAME,sBAAsBC,yBAAyBH,cAAcR;QAEnE,OAAO,IAAI,CAAC,uBAAuB,CAAC,SAAS;YAC3C,QAAQU;QACV;IACF;IAmBA,MAAM,QACJE,mBAAyC,EACzCC,iBAGa,EACbC,cAA6B,EAC7B;QACA,IAAIC;QACJ,IAAIP;QACJ,IAAIR;QAKJ,IACE,AAA6B,YAA7B,OAAOa,qBACPA,AAAsB,SAAtBA,qBACA,WAAWA,mBACX;YAEAL,eAAeI;YACf,MAAMI,eAAeH;YAIrBE,QAAQC,aAAa,KAAK;YAC1BhB,MAAMgB;QACR,OAAO;YAELD,QAAQH;YACRJ,eAAeK;YACfb,MAAM;gBACJ,GAAGc,cAAc;gBACjBC;YACF;QACF;QAEAN,OACE,AAAiB,YAAjB,OAAOM,OACP;QAEFN,OAAOD,cAAc;QAErB,MAAME,sBAAsBC,yBAAyBH,cAAcR;QAEnE,OAAO,IAAI,CAAC,uBAAuB,CAAC,SAAS;YAC3C,GAAIA,OAAO,CAAC,CAAC;YACb,QAAQU;QACV;IACF;IAmBA,MAAM,gBACJO,qBAA2C,EAC3CJ,iBAGa,EACbC,cAA6B,EAC7B;QACA,IAAII;QACJ,IAAIV;QACJ,IAAIR;QAGJ,IACE,AAA6B,YAA7B,OAAOa,qBACPA,AAAsB,SAAtBA,qBACA,aAAaA,mBACb;YAEAL,eAAeS;YACfjB,MAAMa;QAGR,OAAO;YAELK,UAAUD;YACVT,eAAeK;YACfb,MAAM;gBACJ,GAAIc,kBAAkB,CAAC,CAAC;gBACxBI;YACF;QACF;QAEAT,OAAOT,QAAAA,MAAAA,KAAAA,IAAAA,IAAK,OAAO,EAAE;QAErB,MAAMU,sBAAsBF,eACxBG,yBAAyBH,cAAcR,OACvCmB;QAEJ,OAAO,IAAI,CAAC,uBAAuB,CAAC,iBAAiB;YACnD,GAAInB,OAAO,CAAC,CAAC;YACb,QAAQU;QACV;IACF;IAmBA,MAAM,SACJU,yBAAgE,EAChEP,iBAAyE,EACzEC,cAA6B,EAC7B;QACA,IAAIO;QACJ,IAAIb;QACJ,IAAIR;QAGJ,IACE,AAA6B,YAA7B,OAAOa,qBACN,gBAAeA,qBACd,gBAAgBA,qBAChB,cAAcA,iBAAgB,GAChC;YAEAL,eAAeY;YACfpB,MAAMa;QACR,OAAO;YAELQ,cAAcD;YACdZ,eAAeK;YACfb,MAAM;gBACJ,GAAIc,kBAAkB,CAAC,CAAC;gBACxB,GAAIO,eAAe,CAAC,CAAC;YACvB;QACF;QAEA,MAAMX,sBAAsBC,yBAC1BH,gBAAgB,IAChBR;QAGF,OAAO,IAAI,CAAC,uBAAuB,CAAC,UAAU;YAC5C,GAAIA,OAAO,CAAC,CAAC;YACb,QAAQU;QACV;IACF;IAEA,MAAM,SACJY,UAAkB,EAClBtB,GAEC,EACD;YAQMuB,iBACcC;QARpB,MAAMC,mBAAsC;YAAE,QAAQ;QAAW;QACjE,MAAMC,YAAY1B,QAAAA,MAAAA,KAAAA,IAAAA,IAAK,SAAS;QAEhC,MAAM2B,cAAcC,AAAmC,kBAAnCA,aAAaH;QACjC,MAAMI,eACJF,eAAeD,AAAc,UAAdA,YACXP,SAAAA,QACAI,CAAAA,kBAAAA,IAAI,CAAC,SAAS,AAAD,IAAbA,KAAAA,IAAAA,gBAAgB,cAAc,CAACD;QACrC,IAAIO,gBAAAA,SAAgBL,CAAAA,mBAAAA,IAAI,CAAC,SAAS,AAAD,IAAbA,KAAAA,IAAAA,iBAAgB,iBAAiB,AAAD,GAAG;gBAInDM,4BAMWC;YARb,MAAM,EAAEtC,QAAQ,EAAE,GAAG,MAAM,IAAI,CAAC,YAAY,CAAC,sBAAsB,CACjE6B,YAAAA,QACAQ,CAAAA,6BAAAA,aAAa,YAAY,AAAD,IAAxBA,KAAAA,IAAAA,2BAA2B,YAAY;YAGzC,MAAM,MAAM,IAAI,CAAC,gBAAgB,CAACrC;YAElCzC,MAAM;YACN,MAAMgF,OAAO,QAAAD,CAAAA,8BAAAA,aAAa,YAAY,AAAD,IAAxBA,KAAAA,IAAAA,4BAA2B,YAAY;YACpD,OAAO,IAAI,CAAC,OAAO,CAACC;QACtB;QAEA,MAAM,EAAEzB,MAAM,EAAEd,QAAQ,EAAE,GAAG,MAAOkC,CAAAA,cAChC,IAAI,CAAC,YAAY,CAAC,YAAY,CAACL,cAC/B,IAAI,CAAC,YAAY,CAAC,MAAM,CAACA,YAAY,IAAI,CAAC,IAAI,CAAC,eAAe;QAGlE,IAAI,IAAI,CAAC,SAAS,IAAIf,CAAAA,QAAAA,SAAAA,KAAAA,IAAAA,OAAQ,QAAQ,AAAD,KAAKmB,AAAc,UAAdA,WAAqB;YAC7D,MAAMO,cAAkC;gBACtC,OAAO;oBACL;wBACE,MAAMX;wBACN,MAAMf,OAAO,QAAQ;oBACvB;iBACD;YACH;YACA,MAAM2B,cAAcF,QAAAA,IAAS,CAACC;YAC9B,IAAI,CAAC,SAAS,CAAC,yBAAyB,CACtC;gBACE,MAAM;gBACN,QAAQX;gBACR,cAAcY;YAChB,GACAL;QAEJ;QAEA,MAAM,IAAI,CAAC,gBAAgB,CAACpC;QAC5B,OAAOc;IACT;IAEA,MAAM,QACJ4B,MAA2B,EAC3BnC,MAA4B7B,2BAA2B,EAClC;QACrB,MAAM,EAAEoC,MAAM,EAAEd,QAAQ,EAAE,GACxB,MAAM,IAAI,CAAC,YAAY,CAAC,wBAAwB,CAAC,SAAS0C,QAAQnC;QACpE,MAAM,IAAI,CAAC,gBAAgB,CAACP;QAC5B,OAAOc;IACT;IAEA,MAAM,UACJhC,MAAmB,EACnByB,MAA4B7B,2BAA2B,EACrC;QAClB,MAAM,EAAEiE,UAAU,EAAEC,gBAAgB,EAAE,GAAGC,YAAY/D;QACrD,MAAM,EAAEgC,MAAM,EAAEd,QAAQ,EAAE,GACxB,MAAM,IAAI,CAAC,YAAY,CAAC,wBAAwB,CAC9C,WACA2C,YACApC,KACAqC;QAEJ,MAAM,IAAI,CAAC,gBAAgB,CAAC5C;QAC5B,OAAOc;IACT;IAEA,MAAM,SACJhC,MAAmB,EACnByB,MAA4B7B,2BAA2B,EACtC;QACjB,MAAM,EAAEiE,UAAU,EAAEC,gBAAgB,EAAE,GAAGC,YAAY/D;QACrD,MAAM,EAAEgC,MAAM,EAAEd,QAAQ,EAAE,GACxB,MAAM,IAAI,CAAC,YAAY,CAAC,wBAAwB,CAC9C,UACA2C,YACApC,KACAqC;QAEJ,MAAM,IAAI,CAAC,gBAAgB,CAAC5C;QAC5B,OAAOc;IACT;IAEA,MAAM,SACJhC,MAAmB,EACnByB,MAA4B7B,2BAA2B,EACtC;QACjB,MAAM,EAAEiE,UAAU,EAAEC,gBAAgB,EAAE,GAAGC,YAAY/D;QACrD,MAAM,EAAEgC,MAAM,EAAEd,QAAQ,EAAE,GACxB,MAAM,IAAI,CAAC,YAAY,CAAC,wBAAwB,CAC9C,UACA2C,YACApC,KACAqC;QAEJ,MAAM,IAAI,CAAC,gBAAgB,CAAC5C;QAC5B,OAAOc;IACT;IAEA,MAAM,MACJhC,MAAmB,EACnByB,MAA4B7B,2BAA2B,EACtC;QACjB,OAAO,IAAI,CAAC,QAAQ,CAACI,QAAQyB;IAC/B;IAEA,MAAM,uBACJuC,MAAwB,EACxBvC,GAI0B,EACkB;QAC5C,MAAM,EAAEwC,eAAe,IAAI,EAAEC,aAAa,CAAC,EAAE,GAAGzC,OAAO,CAAC;QAExD,IAAI0C,UAAU;QACd,IAAIC,aAAa;QACjB,IAAIC,eAAe;QACnB,IAAIC,YAAY7C,AAAAA,CAAAA,QAAAA,MAAAA,KAAAA,IAAAA,IAAK,SAAS,AAAD,KAAK;QAClC,IAAI8C;QAEJ,MAAO,CAACJ,WAAWC,aAAaF,WAAY;YAC1C,IAAIE,cAAc,GAChBE,YAAY;YAEd7F,MACE,cACAuF,QACA,gBACAC,cACA,cACAG,YACA,aACAE;YAEF,MAAME,OAAO,MAAM,IAAI,CAAC,OAAO,CAAC,QAAQ,CAACR,QAAQ;gBAAEM;YAAU;YAC7D7F,MAAM,mBAAmB+F;YACzBtC,OAAOsC,KAAK,WAAW,EAAE,CAAC,+BAA+B,EAAER,OAAO,CAAC,CAAC;YACpEK,eAAeG,KAAK,WAAW;YAE/BD,eAAe,MAAM,IAAI,CAAC,aAAa,CACrCF,cACAC,YAAY;gBAAE,WAAW;YAAK,IAAI1B,QAClCoB,QACAvC;YAEF,IAAI8C,aAAa,IAAI,EACnBJ,UAAU;iBAEVC;QAEJ;QAEA,OAAO;YACL,QAAQC;YACRC;YACAC;QACF;IACF;IAEA,MAAM,cACJvE,MAAc,EACdyE,SAAmC,EACnCC,YAA8B,EAC9BC,kBAA2C,EACX;QAChClG,MAAM,iBAAiBuB,QAAQyE,WAAWC,cAAcC;QAExD,MAAM,EAAE,QAAQC,YAAY,EAAE,MAAMC,UAAU,EAAE,GAAG,MAAM,IAAI,CAAC,QAAQ,CACpE7E,QACAyE;QAEF,MAAMK,WAAWnG,oBAAoB+F,cAAcE;QACnD,MAAMG,WAAW5F,eAAeuF,cAAcG;QAC9C,MAAMG,OACJF,YAAaH,CAAAA,CAAAA,QAAAA,qBAAAA,KAAAA,IAAAA,mBAAoB,uBAAuB,AAAD,KAAK,EAAC,KAC7DI;QACF,MAAMR,eAAe;YACnBS;YACA,MAAMH;YACN,QAAQD;YACR,gBAAgBE;QAClB;QACArG,MAAM,2BAA2B8F;QACjC,OAAOA;IACT;IAEA,MAAM,SAASvE,MAAmB,EAAEyB,GAAkB,EAAE;QACtD,MAAMwD,cAAc7C,yBAAyBpC,QAAQyB;QACrDS,OAAO+C,aAAa;QACpB,MAAMC,aAAaC,oBAAoBF;QACvC,MAAMtD,QAAQ;YAACuD;SAAW;QAC1B,MAAM,EAAEhE,QAAQ,EAAEc,MAAM,EAAE,GAAG,MAAM,IAAI,CAAC,YAAY,CAAC,QAAQ,CAC3DF,aAAa,UAAUC,eAAekD,eACtCtD;QAEF,MAAM,IAAI,CAAC,gBAAgB,CAACT;QAE5B,MAAM,EAAEkE,OAAO,EAAE,GAAGpD;QAEpB,OAAO;YACL,MAAMoD,QAAAA,UAAAA,KAAAA,IAAAA,QAAS,IAAI;YACnB,QAAQA,QAAAA,UAAAA,KAAAA,IAAAA,QAAS,MAAM;YACvB,OAAQ,OAAM,IAAI,CAAC,SAAS,CAAC,IAAI,EAAC,EAAG,GAAG;QAC1C;IAGF;IAEA,MAAM,SACJC,SAAsB,EACtBC,GAAY,EACZ7D,GAA2C,EAC3C;YAoBiB8D;QAnBjB,MAAMC,aAAmC;YACvC,aAAa/D,AAAAA,CAAAA,QAAAA,MAAAA,KAAAA,IAAAA,IAAK,WAAW,AAAD,KAAK7B,4BAA4B,WAAW;YACxE,oBACE6B,AAAAA,CAAAA,QAAAA,MAAAA,KAAAA,IAAAA,IAAK,kBAAkB,AAAD,KACtB7B,4BAA4B,kBAAkB;YAChD,eAAe6B,AAAAA,CAAAA,QAAAA,MAAAA,KAAAA,IAAAA,IAAK,aAAa,AAAD,KAAK;YACrC,iBAAiBA,QAAAA,MAAAA,KAAAA,IAAAA,IAAK,eAAe;YACrC,iBAAiBA,QAAAA,MAAAA,KAAAA,IAAAA,IAAK,eAAe;QACvC;QAEA,MAAM,EAAEO,MAAM,EAAEd,QAAQ,EAAEuE,OAAO,EAAE,GAAG,MAAM,IAAI,CAAC,YAAY,CAAC,MAAM,CAClEJ,WACAG;QAEF,MAAM,IAAI,CAAC,gBAAgB,CAACtE,UAAU;QAEtC,MAAMwE,UAAU1D,SACZY,SACA,CAAC,kBAAkB,EAAE0C,OAAOD,UAAU,UAAU,EAC9CI,WAAAA,SAAWF,CAAAA,4BAAAA,SAAS,eAAe,EAAC,IAAzBA,KAAAA,IAAAA,0BAA4B,KAAK,AAAD,KAAK,eAChD;QAEN,IAAI9D,QAAAA,MAAAA,KAAAA,IAAAA,IAAK,eAAe,EACtB,OAAO;YACL,MAAMO;YACNyD;YACAC;QACF;QAGF,IAAI,CAAC1D,QACH,MAAM,IAAIzB,MAAMmF;IAEpB;IAEA,MAAM,UAAUL,SAAsB,EAAE5D,GAAqB,EAAE;QAC7D,MAAM,EAAEP,QAAQ,EAAE,GAAG,MAAM,IAAI,CAAC,YAAY,CAAC,OAAO,CAACmE,WAAW;YAC9D,WAAW5D,AAAAA,CAAAA,QAAAA,MAAAA,KAAAA,IAAAA,IAAK,SAAS,AAAD,KAAK;YAC7B,iBAAiBA,AAAAA,CAAAA,QAAAA,MAAAA,KAAAA,IAAAA,IAAK,eAAe,AAAD,KAAK;QAC3C;QACA,MAAM,IAAI,CAAC,gBAAgB,CAACP,UAAU;QAEtC,IAAIA,SAAS,cAAc,IAAI;YAC7B,MAAMK,YAAYL,SAAS,eAAe;YAC1C,MAAM,IAAIX,MAAM,GAAGgB,QAAAA,YAAAA,KAAAA,IAAAA,UAAW,KAAK,CAAC,EAAE,EAAEA,QAAAA,YAAAA,KAAAA,IAAAA,UAAW,UAAU,EAAE;QACjE;IACF;IAEA,MAAM,GAAGwB,UAAkB,EAAEvB,OAAO,QAAQ,EAAE;QAC5C,IAAIA,AAAS,aAATA,MACF,OAAO,IAAI,CAAC,QAAQ,CAACuB;QAEvB,IAAIvB,AAAS,YAATA,MACF,OAAO,IAAI,CAAC,OAAO,CAACuB;QAGtB,IAAIvB,AAAS,aAATA,MACF,OAAO,IAAI,CAAC,QAAQ,CAACuB;QAGvB,IAAIvB,AAAS,UAATA,MACF,OAAO,IAAI,CAAC,KAAK,CAACuB;QAGpB,IAAIvB,AAAS,iBAATA,MACF,OAAO,IAAI,CAAC,YAAY,CAACuB;QAG3B,MAAM,IAAIxC,MACR,CAAC,cAAc,EAAEiB,KAAK,+DAA+D,CAAC;IAE1F;IAEA,MAAM,QAAQmE,iBAAyB,EAEpC;QACD,MAAMC,SAASC,gBAAgBF,mBAAmB,QAAQ;QAC1D,MAAMG,SAAS,IAAIC,aAAaH,QAAQ,OAAOI,SACtC;gBAAE,OAAO,IAAI;gBAAE,QAAQ,EAAE;YAAC;QAEnC,MAAMF,OAAO,GAAG;QAEhB,IAAIA,AAAkB,YAAlBA,OAAO,MAAM,EAAc;YAC7B,MAAMG,SAASH,OAAO,cAAc,CACjC,MAAM,CAAC,CAACjF,OAASA,AAAgB,YAAhBA,KAAK,MAAM,EAC5B,GAAG,CAAC,CAACA;oBAC2BqF;gBAA/B,OAAO,CAAC,OAAO,EAAErF,KAAK,IAAI,CAAC,EAAE,EAAE,QAAAqF,CAAAA,cAAAA,KAAK,KAAK,AAAD,IAATA,KAAAA,IAAAA,YAAY,OAAO,EAAE;YACtD,GACC,IAAI,CAAC;YACR,MAAM,IAAI3F,MAAM,CAAC,2CAA2C,EAAE0F,QAAQ;QACxE;QAEA,OAAO;YACL,QAAQH,OAAO,MAAM;QACvB;IACF;IAEA,MAAM,mBAAmBF,MAAc,EAAE;QACvC1D,OACE,IAAI,CAAC,SAAS,CAAC,kBAAkB,EACjC;QAEF,OAAO,IAAI,CAAC,SAAS,CAAC,kBAAkB,CAAC0D;IAC3C;IAEA,MAAM,UAAU;QACd,MAAM,IAAI,CAAC,SAAS,CAAC,OAAO;QAC5B,IAAI,CAAC,SAAS;QACd,IAAI,CAAC,SAAS,GAAG;IACnB;IAEA,MAAM,cACJ/D,KAAc,EACdJ,GAEC,EACD;QAEA,MAAM0E,SAAS,MAAM,IAAI,CAAC,SAAS,CAAC,gBAAgB;QACpD,MAAMC,MAAMC,KAAK,GAAG;QAEpB,MAAMC,WAAoC;YACxC;gBACE,MAAM;gBACN,IAAIF;gBACJ,YAAYD;YACd;SACD;QAED,MAAMtF,OAAyB;YAC7B,MAAM;YACN,SAAS;YACT,QAAQ;YACRyF;YACA,QAAQ;gBACN,OAAOF;gBACP,KAAKA;gBACL,MAAM;YACR;YACA,OAAO;gBACL,SAAS3E,AAAAA,CAAAA,QAAAA,MAAAA,KAAAA,IAAAA,IAAK,OAAO,AAAD,KAAK;YAC3B;YACA,UAAU,WAAa;QACzB;QAEA,MAAM8E,gBAA+B;YACnC,YAAY;YACZ,SAASH;YACT,MAAM,CAAC,MAAM,EAAEvE,SAAS,YAAY;YACpC,aAAaJ,AAAAA,CAAAA,QAAAA,MAAAA,KAAAA,IAAAA,IAAK,OAAO,AAAD,KAAK;YAC7B,OAAO;gBAACZ;aAAK;QACf;QAEA,IAAI,CAAC,mBAAmB,CAAC0F;QAEzB,IAAI;gBACFnF,oBAAAA;oBAAAA,CAAAA,qBAAAA,AAAAA,CAAAA,QAAAA,IAAI,AAAD,EAAE,YAAY,AAAD,KAAhBA,mBAAAA,IAAAA,CAAAA,OAAoB,IAAI,CAAC,cAAc;QACzC,EAAE,OAAOC,OAAO;YACdC,QAAQ,KAAK,CAAC,yBAAyBD;QACzC;QAEA,IAAI,CAAC,mBAAmB;IAC1B;IAEA,sBAAsB;QACpB,MAAM,EAAEmF,SAAS,EAAEC,gBAAgB,EAAEC,UAAU,EAAE,GAAG,IAAI,CAAC,IAAI;QAC7D,MAAMC,gBAAgBC,MAAM,OAAO,CAACF,cAChCA,WAAW,GAAG,CAAC,CAACzG;YACd,MAAM,EAAE4G,KAAK,EAAE,GAAGC,eAAe,GAAG7G;YACpC,IAAI8G,WAAWF;YACf,IAAID,MAAM,OAAO,CAACC,QAChBE,WAAWF,MAAM,GAAG,CAAC,CAAChG;gBAEpB,MAAM,EAAEmG,SAAS,EAAEC,GAAG,EAAE,GAAGC,UAAU,GAAGrG;gBACxC,OAAOqG;YACT;YAEF,OAAO;gBAAE,GAAGJ,aAAa;gBAAE,GAAIC,WAAW;oBAAE,OAAOA;gBAAS,IAAI,CAAC,CAAC;YAAE;QACtE,KACA,EAAE;QACN,OAAO;YACLP;YACAC;YACA,YAAYE;QACd;IACF;IAMA,MAAM,oBAAmC;QACvClI,MAAM;QACN,MAAM0I,UAAU,MAAM,IAAI,CAAC,gBAAgB;QAE3CA,QAAQ,SAAS,GAAG;QACpB,IAAI,CAAC,eAAe,GAAGA;QACvB1I,MAAM;IACR;IAKA,MAAM,sBAAqC;QACzCA,MAAM;QACN,IAAI,CAAC,eAAe,GAAGmE;QACvBnE,MAAM;IACR;IA9zBA,YAAY2I,iBAAgC,EAAEC,IAAe,CAAE;QArC/D;QAEA;QAEA;QAEA;QAEA;QAEA;QAEA;QAKA,kCAAU;QAEV;QAEA;QAEA;QAEA,oCAAY;QAKZ,uBAAQ,mBAAR;QAQE,IAAI,CAAC,SAAS,GAAGD;QACjB,IAAI,CAAC,IAAI,GAAGE,OAAO,MAAM,CACvB;YACE,gBAAgB;YAChB,oBAAoB;YACpB,WAAW;YACX,kBAAkB;QACpB,GACAD,QAAQ,CAAC;QAEX,IAAIA,AAAAA,CAAAA,QAAAA,OAAAA,KAAAA,IAAAA,KAAM,WAAW,AAAD,KAAK,AAA6B,cAA7B,OAAOA,CAAAA,QAAAA,OAAAA,KAAAA,IAAAA,KAAM,WAAW,AAAD,GAC9C,MAAM,IAAI9G,MACR,CAAC,+DAA+D,EAAE,OAAO8G,CAAAA,QAAAA,OAAAA,KAAAA,IAAAA,KAAM,WAAW,AAAD,GAAG;QAIhGE,oBAAoB,IAAI,CAACF,QAAAA,OAAAA,KAAAA,IAAAA,KAAM,WAAW;QAE1C,IAAI,CAAC,cAAc,GAAG,IAAI,CAAC,IAAI,CAAC,cAAc;QAE9C,IAAI,CAAC,OAAO,GAAG,IAAIG,QAAQ,OAAO1H,SACzB,IAAI,CAAC,YAAY,CAACA;QAG3B,IAAIuH,AAAAA,CAAAA,QAAAA,OAAAA,KAAAA,IAAAA,KAAM,OAAO,AAAD,KAAK,AAAiC,cAAjC,IAAI,CAAC,SAAS,CAAC,aAAa,EAC/C,IAAI,CAAC,SAAS,GAAG,IAAII,UACnBJ,KAAK,OAAO,EACZE,oBAAoB,qBAAqB,CAACG;QAI9C,IAAI,CAAC,YAAY,GAAG,IAAIC,aAAa,IAAI,CAAC,SAAS,EAAE,IAAI,CAAC,OAAO,EAAE;YACjE,WAAW,IAAI,CAAC,SAAS;YACzB,aAAa,IAAI,CAAC,sBAAsB,CAAC,IAAI,CAAC,IAAI;QACpD;QACA,IAAI,CAAC,IAAI,GAAG,IAAI,CAAC,SAAS;QAC1B,IAAI,CAAC,cAAc,GACjBN,AAAAA,CAAAA,QAAAA,OAAAA,KAAAA,IAAAA,KAAM,cAAc,AAAD,KACnBO,kBAAkBP,AAAAA,CAAAA,QAAAA,OAAAA,KAAAA,IAAAA,KAAM,MAAM,AAAD,KAAK,IAAI,CAAC,SAAS,CAAC,aAAa,IAAI;IACtE;AAuxBF"}
@@ -2,5 +2,5 @@ import { Agent } from "./agent.mjs";
2
2
  import { commonContextParser, getCurrentExecutionFile, getReportFileName, printReportMsg, trimContextByViewport } from "./utils.mjs";
3
3
  import { locateParamStr, paramStr, taskTitleStr, typeStr } from "./ui-utils.mjs";
4
4
  import { TaskCache, cacheFileExt } from "./task-cache.mjs";
5
- import { PageTaskExecutor } from "./tasks.mjs";
6
- export { Agent, PageTaskExecutor, TaskCache, cacheFileExt, commonContextParser, getCurrentExecutionFile, getReportFileName, locateParamStr, paramStr, printReportMsg, taskTitleStr, trimContextByViewport, typeStr };
5
+ import { TaskExecutor } from "./tasks.mjs";
6
+ export { Agent, TaskCache, TaskExecutor, cacheFileExt, commonContextParser, getCurrentExecutionFile, getReportFileName, locateParamStr, paramStr, printReportMsg, taskTitleStr, trimContextByViewport, typeStr };
@@ -31,9 +31,12 @@ function locatePlanForLocate(param) {
31
31
  };
32
32
  return locatePlan;
33
33
  }
34
- class PageTaskExecutor {
34
+ class TaskExecutor {
35
+ get page() {
36
+ return this.interface;
37
+ }
35
38
  async recordScreenshot(timing) {
36
- const base64 = await this.page.screenshotBase64();
39
+ const base64 = await this.interface.screenshotBase64();
37
40
  const item = {
38
41
  type: 'screenshot',
39
42
  ts: Date.now(),
@@ -42,12 +45,12 @@ class PageTaskExecutor {
42
45
  };
43
46
  return item;
44
47
  }
45
- async getElementXpath(pageContext, element) {
48
+ async getElementXpath(uiContext, element) {
46
49
  var _element_attributes;
47
- if (!this.page.getXpathsByPoint) return void debug('getXpathsByPoint is not supported for this page');
50
+ if (!this.interface.getXpathsByPoint) return void debug('getXpathsByPoint is not supported for this interface');
48
51
  let elementId = null == element ? void 0 : element.id;
49
52
  if ((null == element ? void 0 : element.isOrderSensitive) !== void 0) {
50
- const xpaths = await this.page.getXpathsByPoint({
53
+ const xpaths = await this.interface.getXpathsByPoint({
51
54
  left: element.center[0],
52
55
  top: element.center[1]
53
56
  }, null == element ? void 0 : element.isOrderSensitive);
@@ -55,7 +58,7 @@ class PageTaskExecutor {
55
58
  }
56
59
  if ((null == element ? void 0 : null == (_element_attributes = element.attributes) ? void 0 : _element_attributes.nodeType) === NodeType.POSITION) {
57
60
  await this.insight.contextRetrieverFn('locate');
58
- const info = elementByPositionWithElementInfo(pageContext.tree, {
61
+ const info = elementByPositionWithElementInfo(uiContext.tree, {
59
62
  x: element.center[0],
60
63
  y: element.center[1]
61
64
  }, {
@@ -67,7 +70,7 @@ class PageTaskExecutor {
67
70
  }
68
71
  if (!elementId) return;
69
72
  try {
70
- const result = await this.page.getXpathsById(elementId);
73
+ const result = await this.interface.getXpathsById(elementId);
71
74
  return result;
72
75
  } catch (error) {
73
76
  debug('getXpathsById error: ', error);
@@ -86,9 +89,9 @@ class PageTaskExecutor {
86
89
  if ('Action' === taskApply.type) await Promise.all([
87
90
  (async ()=>{
88
91
  await external_utils_mjs_sleep(100);
89
- if (this.page.beforeAction) {
90
- debug('will call "beforeAction" for page');
91
- await this.page.beforeAction();
92
+ if (this.interface.beforeAction) {
93
+ debug('will call "beforeAction" for interface');
94
+ await this.interface.beforeAction();
92
95
  }
93
96
  })(),
94
97
  external_utils_mjs_sleep(200)
@@ -130,34 +133,34 @@ class PageTaskExecutor {
130
133
  };
131
134
  this.insight.onceDumpUpdatedFn = dumpCollector;
132
135
  const shotTime = Date.now();
133
- const pageContext = await this.insight.contextRetrieverFn('locate');
134
- task.pageContext = pageContext;
136
+ const uiContext = await this.insight.contextRetrieverFn('locate');
137
+ task.uiContext = uiContext;
135
138
  const recordItem = {
136
139
  type: 'screenshot',
137
140
  ts: shotTime,
138
- screenshot: pageContext.screenshotBase64,
141
+ screenshot: uiContext.screenshotBase64,
139
142
  timing: 'before Insight'
140
143
  };
141
144
  task.recorder = [
142
145
  recordItem
143
146
  ];
144
- const elementFromXpath = param.xpath && this.page.getElementInfoByXpath ? await this.page.getElementInfoByXpath(param.xpath) : void 0;
147
+ const elementFromXpath = param.xpath && this.interface.getElementInfoByXpath ? await this.interface.getElementInfoByXpath(param.xpath) : void 0;
145
148
  const userExpectedPathHitFlag = !!elementFromXpath;
146
149
  const cachePrompt = param.prompt;
147
150
  const locateCacheRecord = null == (_this_taskCache = this.taskCache) ? void 0 : _this_taskCache.matchLocateCache(cachePrompt);
148
151
  const xpaths = null == locateCacheRecord ? void 0 : null == (_locateCacheRecord_cacheContent = locateCacheRecord.cacheContent) ? void 0 : _locateCacheRecord_cacheContent.xpaths;
149
152
  const elementFromCache = userExpectedPathHitFlag ? null : await matchElementFromCache(this, xpaths, cachePrompt, param.cacheable);
150
153
  const cacheHitFlag = !!elementFromCache;
151
- const elementFromPlan = userExpectedPathHitFlag || cacheHitFlag ? void 0 : matchElementFromPlan(param, pageContext.tree);
154
+ const elementFromPlan = userExpectedPathHitFlag || cacheHitFlag ? void 0 : matchElementFromPlan(param, uiContext.tree);
152
155
  const planHitFlag = !!elementFromPlan;
153
156
  const elementFromAiLocate = userExpectedPathHitFlag || cacheHitFlag || planHitFlag ? void 0 : (await this.insight.locate(param, {
154
- context: pageContext
157
+ context: uiContext
155
158
  })).element;
156
159
  const aiLocateHitFlag = !!elementFromAiLocate;
157
160
  const element = elementFromXpath || elementFromCache || elementFromPlan || elementFromAiLocate;
158
161
  let currentXpaths;
159
162
  if (element && this.taskCache && !cacheHitFlag && (null == param ? void 0 : param.cacheable) !== false) {
160
- const elementXpaths = await this.getElementXpath(pageContext, element);
163
+ const elementXpaths = await this.getElementXpath(uiContext, element);
161
164
  if (null == elementXpaths ? void 0 : elementXpaths.length) {
162
165
  currentXpaths = elementXpaths;
163
166
  this.taskCache.updateOrAppendCacheRecord({
@@ -200,7 +203,7 @@ class PageTaskExecutor {
200
203
  output: {
201
204
  element
202
205
  },
203
- pageContext,
206
+ uiContext,
204
207
  hitBy
205
208
  };
206
209
  }
@@ -253,7 +256,7 @@ class PageTaskExecutor {
253
256
  tasks.push(taskActionSleep);
254
257
  } else {
255
258
  const planType = plan.type;
256
- const actionSpace = await this.page.actionSpace();
259
+ const actionSpace = await this.interface.actionSpace();
257
260
  const action = actionSpace.find((action)=>action.name === planType);
258
261
  const param = plan.param;
259
262
  if (!action) throw new Error(`Action type '${planType}' not found`);
@@ -280,12 +283,12 @@ class PageTaskExecutor {
280
283
  executor: async (param, context)=>{
281
284
  var _context_element;
282
285
  debug('executing action', planType, param, `context.element.center: ${null == (_context_element = context.element) ? void 0 : _context_element.center}`);
283
- const pageContext = await this.insight.contextRetrieverFn('locate');
284
- context.task.pageContext = pageContext;
286
+ const uiContext = await this.insight.contextRetrieverFn('locate');
287
+ context.task.uiContext = uiContext;
285
288
  requiredLocateFields.forEach((field)=>{
286
289
  assert(param[field], `field '${field}' is required for action ${planType} but not provided. Cannot execute action ${planType}.`);
287
290
  });
288
- const actionFn = action.call.bind(this.page);
291
+ const actionFn = action.call.bind(this.interface);
289
292
  await actionFn(param, context);
290
293
  return {
291
294
  output: {
@@ -308,19 +311,19 @@ class PageTaskExecutor {
308
311
  }
309
312
  async setupPlanningContext(executorContext) {
310
313
  const shotTime = Date.now();
311
- const pageContext = await this.insight.contextRetrieverFn('locate');
314
+ const uiContext = await this.insight.contextRetrieverFn('locate');
312
315
  const recordItem = {
313
316
  type: 'screenshot',
314
317
  ts: shotTime,
315
- screenshot: pageContext.screenshotBase64,
318
+ screenshot: uiContext.screenshotBase64,
316
319
  timing: 'before Planning'
317
320
  };
318
321
  executorContext.task.recorder = [
319
322
  recordItem
320
323
  ];
321
- executorContext.task.pageContext = pageContext;
324
+ executorContext.task.uiContext = uiContext;
322
325
  return {
323
- pageContext
326
+ uiContext
324
327
  };
325
328
  }
326
329
  async loadYamlFlowAsPlanning(userInstruction, yamlString) {
@@ -372,17 +375,17 @@ class PageTaskExecutor {
372
375
  },
373
376
  executor: async (param, executorContext)=>{
374
377
  const startTime = Date.now();
375
- const { pageContext } = await this.setupPlanningContext(executorContext);
376
- assert(this.page.actionSpace, 'actionSpace for device is not implemented');
377
- const actionSpace = await this.page.actionSpace();
378
- debug('actionSpace for page is:', actionSpace.map((action)=>action.name).join(', '));
378
+ const { uiContext } = await this.setupPlanningContext(executorContext);
379
+ assert(this.interface.actionSpace, 'actionSpace for device is not implemented');
380
+ const actionSpace = await this.interface.actionSpace();
381
+ debug('actionSpace for this interface is:', actionSpace.map((action)=>action.name).join(', '));
379
382
  assert(Array.isArray(actionSpace), 'actionSpace must be an array');
380
- if (0 === actionSpace.length) console.warn(`ActionSpace for ${this.page.pageType} is empty. This may lead to unexpected behavior.`);
383
+ if (0 === actionSpace.length) console.warn(`ActionSpace for ${this.interface.interfaceType} is empty. This may lead to unexpected behavior.`);
381
384
  const planResult = await external_index_mjs_plan(param.userInstruction, {
382
- context: pageContext,
385
+ context: uiContext,
383
386
  log: param.log,
384
387
  actionContext,
385
- pageType: this.page.pageType,
388
+ interfaceType: this.interface.interfaceType,
386
389
  actionSpace
387
390
  });
388
391
  const { actions, log, more_actions_needed_by_instruction, error, usage, rawResponse, sleep } = planResult;
@@ -414,7 +417,7 @@ class PageTaskExecutor {
414
417
  cache: {
415
418
  hit: false
416
419
  },
417
- pageContext
420
+ uiContext
418
421
  };
419
422
  }
420
423
  };
@@ -430,8 +433,8 @@ class PageTaskExecutor {
430
433
  },
431
434
  executor: async (param, executorContext)=>{
432
435
  var _actions_;
433
- const { pageContext } = await this.setupPlanningContext(executorContext);
434
- const imagePayload = await resizeImageForUiTars(pageContext.screenshotBase64, pageContext.size, modelPreferences);
436
+ const { uiContext } = await this.setupPlanningContext(executorContext);
437
+ const imagePayload = await resizeImageForUiTars(uiContext.screenshotBase64, uiContext.size, modelPreferences);
435
438
  this.appendConversationHistory({
436
439
  role: 'user',
437
440
  content: [
@@ -446,7 +449,7 @@ class PageTaskExecutor {
446
449
  const planResult = await vlmPlanning({
447
450
  userInstruction: param.userInstruction,
448
451
  conversationHistory: this.conversationHistory,
449
- size: pageContext.size,
452
+ size: uiContext.size,
450
453
  modelPreferences
451
454
  });
452
455
  const { actions, action_summary, usage } = planResult;
@@ -604,12 +607,12 @@ class PageTaskExecutor {
604
607
  };
605
608
  this.insight.onceDumpUpdatedFn = dumpCollector;
606
609
  const shotTime = Date.now();
607
- const pageContext = await this.insight.contextRetrieverFn('extract');
608
- task.pageContext = pageContext;
610
+ const uiContext = await this.insight.contextRetrieverFn('extract');
611
+ task.uiContext = uiContext;
609
612
  const recordItem = {
610
613
  type: 'screenshot',
611
614
  ts: shotTime,
612
- screenshot: pageContext.screenshotBase64,
615
+ screenshot: uiContext.screenshotBase64,
613
616
  timing: 'before Extract'
614
617
  };
615
618
  task.recorder = [
@@ -737,18 +740,18 @@ class PageTaskExecutor {
737
740
  }
738
741
  return this.appendErrorPlan(taskExecutor, `waitFor timeout: ${errorThought}`);
739
742
  }
740
- constructor(page, insight, opts){
741
- _define_property(this, "page", void 0);
743
+ constructor(interfaceInstance, insight, opts){
744
+ _define_property(this, "interface", void 0);
742
745
  _define_property(this, "insight", void 0);
743
746
  _define_property(this, "taskCache", void 0);
744
747
  _define_property(this, "conversationHistory", []);
745
748
  _define_property(this, "onTaskStartCallback", void 0);
746
- this.page = page;
749
+ this.interface = interfaceInstance;
747
750
  this.insight = insight;
748
751
  this.taskCache = opts.taskCache;
749
752
  this.onTaskStartCallback = null == opts ? void 0 : opts.onTaskStart;
750
753
  }
751
754
  }
752
- export { PageTaskExecutor, locatePlanForLocate };
755
+ export { TaskExecutor, locatePlanForLocate };
753
756
 
754
757
  //# sourceMappingURL=tasks.mjs.map