@midscene/core 0.28.2 → 0.28.3-beta-20250910131004.0

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 (85) hide show
  1. package/dist/es/agent/agent.mjs +27 -17
  2. package/dist/es/agent/agent.mjs.map +1 -1
  3. package/dist/es/agent/tasks.mjs +43 -31
  4. package/dist/es/agent/tasks.mjs.map +1 -1
  5. package/dist/es/agent/utils.mjs +4 -3
  6. package/dist/es/agent/utils.mjs.map +1 -1
  7. package/dist/es/ai-model/common.mjs +12 -22
  8. package/dist/es/ai-model/common.mjs.map +1 -1
  9. package/dist/es/ai-model/index.mjs +3 -3
  10. package/dist/es/ai-model/inspect.mjs +26 -34
  11. package/dist/es/ai-model/inspect.mjs.map +1 -1
  12. package/dist/es/ai-model/llm-planning.mjs +14 -15
  13. package/dist/es/ai-model/llm-planning.mjs.map +1 -1
  14. package/dist/es/ai-model/prompt/common.mjs.map +1 -1
  15. package/dist/es/ai-model/prompt/llm-locator.mjs.map +1 -1
  16. package/dist/es/ai-model/prompt/llm-planning.mjs.map +1 -1
  17. package/dist/es/ai-model/prompt/llm-section-locator.mjs.map +1 -1
  18. package/dist/es/ai-model/prompt/playwright-generator.mjs +6 -12
  19. package/dist/es/ai-model/prompt/playwright-generator.mjs.map +1 -1
  20. package/dist/es/ai-model/prompt/util.mjs +3 -3
  21. package/dist/es/ai-model/prompt/util.mjs.map +1 -1
  22. package/dist/es/ai-model/prompt/yaml-generator.mjs +6 -12
  23. package/dist/es/ai-model/prompt/yaml-generator.mjs.map +1 -1
  24. package/dist/es/ai-model/service-caller/index.mjs +28 -23
  25. package/dist/es/ai-model/service-caller/index.mjs.map +1 -1
  26. package/dist/es/ai-model/ui-tars-planning.mjs +10 -10
  27. package/dist/es/ai-model/ui-tars-planning.mjs.map +1 -1
  28. package/dist/es/insight/index.mjs +20 -26
  29. package/dist/es/insight/index.mjs.map +1 -1
  30. package/dist/es/types.mjs.map +1 -1
  31. package/dist/es/utils.mjs +4 -5
  32. package/dist/es/utils.mjs.map +1 -1
  33. package/dist/lib/agent/agent.js +26 -16
  34. package/dist/lib/agent/agent.js.map +1 -1
  35. package/dist/lib/agent/tasks.js +43 -31
  36. package/dist/lib/agent/tasks.js.map +1 -1
  37. package/dist/lib/agent/utils.js +4 -3
  38. package/dist/lib/agent/utils.js.map +1 -1
  39. package/dist/lib/ai-model/common.js +14 -27
  40. package/dist/lib/ai-model/common.js.map +1 -1
  41. package/dist/lib/ai-model/index.js +12 -15
  42. package/dist/lib/ai-model/inspect.js +24 -32
  43. package/dist/lib/ai-model/inspect.js.map +1 -1
  44. package/dist/lib/ai-model/llm-planning.js +13 -14
  45. package/dist/lib/ai-model/llm-planning.js.map +1 -1
  46. package/dist/lib/ai-model/prompt/common.js.map +1 -1
  47. package/dist/lib/ai-model/prompt/llm-locator.js.map +1 -1
  48. package/dist/lib/ai-model/prompt/llm-planning.js.map +1 -1
  49. package/dist/lib/ai-model/prompt/llm-section-locator.js.map +1 -1
  50. package/dist/lib/ai-model/prompt/playwright-generator.js +5 -11
  51. package/dist/lib/ai-model/prompt/playwright-generator.js.map +1 -1
  52. package/dist/lib/ai-model/prompt/util.js +3 -3
  53. package/dist/lib/ai-model/prompt/util.js.map +1 -1
  54. package/dist/lib/ai-model/prompt/yaml-generator.js +5 -11
  55. package/dist/lib/ai-model/prompt/yaml-generator.js.map +1 -1
  56. package/dist/lib/ai-model/service-caller/index.js +36 -31
  57. package/dist/lib/ai-model/service-caller/index.js.map +1 -1
  58. package/dist/lib/ai-model/ui-tars-planning.js +8 -8
  59. package/dist/lib/ai-model/ui-tars-planning.js.map +1 -1
  60. package/dist/lib/insight/index.js +17 -23
  61. package/dist/lib/insight/index.js.map +1 -1
  62. package/dist/lib/types.js.map +1 -1
  63. package/dist/lib/utils.js +3 -4
  64. package/dist/lib/utils.js.map +1 -1
  65. package/dist/types/agent/agent.d.ts +2 -0
  66. package/dist/types/agent/tasks.d.ts +8 -7
  67. package/dist/types/agent/utils.d.ts +3 -1
  68. package/dist/types/ai-model/common.d.ts +9 -16
  69. package/dist/types/ai-model/index.d.ts +2 -2
  70. package/dist/types/ai-model/inspect.d.ts +7 -6
  71. package/dist/types/ai-model/llm-planning.d.ts +2 -2
  72. package/dist/types/ai-model/prompt/common.d.ts +2 -2
  73. package/dist/types/ai-model/prompt/llm-locator.d.ts +2 -2
  74. package/dist/types/ai-model/prompt/llm-planning.d.ts +3 -3
  75. package/dist/types/ai-model/prompt/llm-section-locator.d.ts +2 -2
  76. package/dist/types/ai-model/prompt/playwright-generator.d.ts +3 -2
  77. package/dist/types/ai-model/prompt/util.d.ts +3 -2
  78. package/dist/types/ai-model/prompt/yaml-generator.d.ts +3 -2
  79. package/dist/types/ai-model/service-caller/index.d.ts +5 -5
  80. package/dist/types/ai-model/ui-tars-planning.d.ts +3 -3
  81. package/dist/types/device/index.d.ts +2 -2
  82. package/dist/types/insight/index.d.ts +12 -7
  83. package/dist/types/types.d.ts +0 -9
  84. package/dist/types/utils.d.ts +2 -1
  85. package/package.json +3 -3
@@ -1 +1 @@
1
- {"version":3,"file":"agent/utils.js","sources":["webpack://@midscene/core/webpack/runtime/compat_get_default_export","webpack://@midscene/core/webpack/runtime/define_property_getters","webpack://@midscene/core/webpack/runtime/has_own_property","webpack://@midscene/core/webpack/runtime/make_namespace_object","webpack://@midscene/core/./src/agent/utils.ts"],"sourcesContent":["// getDefaultExport function for compatibility with non-ESM modules\n__webpack_require__.n = (module) => {\n\tvar getter = module && module.__esModule ?\n\t\t() => (module['default']) :\n\t\t() => (module);\n\t__webpack_require__.d(getter, { a: getter });\n\treturn getter;\n};\n","__webpack_require__.d = (exports, definition) => {\n\tfor(var key in definition) {\n if(__webpack_require__.o(definition, key) && !__webpack_require__.o(exports, key)) {\n Object.defineProperty(exports, key, { enumerable: true, get: definition[key] });\n }\n }\n};","__webpack_require__.o = (obj, prop) => (Object.prototype.hasOwnProperty.call(obj, prop))","// define __esModule on exports\n__webpack_require__.r = (exports) => {\n\tif(typeof Symbol !== 'undefined' && Symbol.toStringTag) {\n\t\tObject.defineProperty(exports, Symbol.toStringTag, { value: 'Module' });\n\t}\n\tObject.defineProperty(exports, '__esModule', { value: true });\n};","import { elementByPositionWithElementInfo } from '@/ai-model';\nimport type { AbstractInterface } from '@/device';\nimport type {\n BaseElement,\n ElementTreeNode,\n ExecutionDump,\n ExecutionTask,\n ExecutorContext,\n PlanningLocateParam,\n TMultimodalPrompt,\n TUserPrompt,\n UIContext,\n} from '@/index';\nimport { uploadTestInfoToServer } from '@/utils';\nimport {\n MIDSCENE_REPORT_TAG_NAME,\n globalConfigManager,\n} from '@midscene/shared/env';\nimport {\n generateElementByPosition,\n getNodeFromCacheList,\n} from '@midscene/shared/extractor';\nimport { resizeImgBase64 } from '@midscene/shared/img';\nimport { getDebug } from '@midscene/shared/logger';\nimport { _keyDefinitions } from '@midscene/shared/us-keyboard-layout';\nimport { assert, logMsg, uuid } from '@midscene/shared/utils';\nimport dayjs from 'dayjs';\nimport { debug as cacheDebug } from './task-cache';\nimport type { TaskExecutor } from './tasks';\n\nconst debugProfile = getDebug('web:tool:profile');\n\nexport async function commonContextParser(\n interfaceInstance: AbstractInterface,\n): Promise<UIContext> {\n assert(interfaceInstance, 'interfaceInstance is required');\n\n debugProfile('Getting interface description');\n const description = interfaceInstance.describe?.() || '';\n debugProfile('Interface description end');\n\n debugProfile('Uploading test info to server');\n uploadTestInfoToServer({ testUrl: description });\n debugProfile('UploadTestInfoToServer end');\n\n let screenshotBase64 = await interfaceInstance.screenshotBase64();\n assert(screenshotBase64!, 'screenshotBase64 is required');\n\n const size = await interfaceInstance.size();\n debugProfile(`size: ${size.width}x${size.height} dpr: ${size.dpr}`);\n\n if (size.dpr && size.dpr > 1) {\n debugProfile('Resizing screenshot for high DPR display');\n screenshotBase64 = await resizeImgBase64(screenshotBase64, {\n width: size.width,\n height: size.height,\n });\n debugProfile('ResizeImgBase64 end');\n }\n\n return {\n tree: {\n node: null,\n children: [],\n },\n size,\n screenshotBase64: screenshotBase64!,\n };\n}\n\nexport function getReportFileName(tag = 'web') {\n const reportTagName = globalConfigManager.getEnvConfigValue(\n MIDSCENE_REPORT_TAG_NAME,\n );\n const dateTimeInFileName = dayjs().format('YYYY-MM-DD_HH-mm-ss');\n // ensure uniqueness at the same time\n const uniqueId = uuid().substring(0, 8);\n return `${reportTagName || tag}-${dateTimeInFileName}-${uniqueId}`;\n}\n\nexport function printReportMsg(filepath: string) {\n logMsg(`Midscene - report file updated: ${filepath}`);\n}\n\n/**\n * Get the current execution file name\n * @returns The name of the current execution file\n */\nexport function getCurrentExecutionFile(trace?: string): string | false {\n const error = new Error();\n const stackTrace = trace || error.stack;\n const pkgDir = process.cwd() || '';\n if (stackTrace) {\n const stackLines = stackTrace.split('\\n');\n for (const line of stackLines) {\n if (\n line.includes('.spec.') ||\n line.includes('.test.') ||\n line.includes('.ts') ||\n line.includes('.js')\n ) {\n const match = line.match(/(?:at\\s+)?(.*?\\.(?:spec|test)\\.[jt]s)/);\n if (match?.[1]) {\n const targetFileName = match[1]\n .replace(pkgDir, '')\n .trim()\n .replace('at ', '');\n return targetFileName;\n }\n }\n }\n }\n return false;\n}\n\nconst testFileIndex = new Map<string, number>();\n\nexport function generateCacheId(fileName?: string): string {\n let taskFile = fileName || getCurrentExecutionFile();\n if (!taskFile) {\n taskFile = uuid();\n console.warn(\n 'Midscene - using random UUID for cache id. Cache may be invalid.',\n );\n }\n\n if (testFileIndex.has(taskFile)) {\n const currentIndex = testFileIndex.get(taskFile);\n if (currentIndex !== undefined) {\n testFileIndex.set(taskFile, currentIndex + 1);\n }\n } else {\n testFileIndex.set(taskFile, 1);\n }\n return `${taskFile}-${testFileIndex.get(taskFile)}`;\n}\n\nexport function matchElementFromPlan(\n planLocateParam: PlanningLocateParam,\n tree: ElementTreeNode<BaseElement>,\n) {\n if (!planLocateParam) {\n return undefined;\n }\n if (planLocateParam.id) {\n return getNodeFromCacheList(planLocateParam.id);\n }\n\n if (planLocateParam.bbox) {\n const centerPosition = {\n x: Math.floor((planLocateParam.bbox[0] + planLocateParam.bbox[2]) / 2),\n y: Math.floor((planLocateParam.bbox[1] + planLocateParam.bbox[3]) / 2),\n };\n let element = elementByPositionWithElementInfo(tree, centerPosition);\n\n if (!element) {\n element = generateElementByPosition(centerPosition) as BaseElement;\n }\n\n return element;\n }\n\n return undefined;\n}\n\nexport async function matchElementFromCache(\n taskExecutor: TaskExecutor,\n xpaths: string[] | undefined,\n cachePrompt: TUserPrompt,\n cacheable: boolean | undefined,\n) {\n try {\n if (\n xpaths?.length &&\n taskExecutor.taskCache?.isCacheResultUsed &&\n cacheable !== false &&\n (taskExecutor.interface as any).getElementInfoByXpath\n ) {\n // hit cache, use new id\n for (let i = 0; i < xpaths.length; i++) {\n const element = await (\n taskExecutor.interface as any\n ).getElementInfoByXpath(xpaths[i]);\n\n if (element?.id) {\n cacheDebug('cache hit, prompt: %s', cachePrompt);\n cacheDebug(\n 'found a new element with same xpath, xpath: %s, id: %s',\n xpaths[i],\n element?.id,\n );\n return element;\n }\n }\n }\n } catch (error) {\n cacheDebug('get element info by xpath error: ', error);\n }\n}\n\nexport function trimContextByViewport(execution: ExecutionDump) {\n function filterVisibleTree(\n node: ElementTreeNode<BaseElement>,\n ): ElementTreeNode<BaseElement> | null {\n if (!node) return null;\n\n // recursively process all children\n const filteredChildren = Array.isArray(node.children)\n ? (node.children\n .map(filterVisibleTree)\n .filter((child) => child !== null) as ElementTreeNode<BaseElement>[])\n : [];\n\n // if the current node is visible, keep it and the filtered children\n if (node.node && node.node.isVisible === true) {\n return {\n ...node,\n children: filteredChildren,\n };\n }\n\n // if the current node is invisible, but has visible children, create an empty node to include these children\n if (filteredChildren.length > 0) {\n return {\n node: null,\n children: filteredChildren,\n };\n }\n\n // if the current node is invisible and has no visible children, return null\n return null;\n }\n\n return {\n ...execution,\n tasks: Array.isArray(execution.tasks)\n ? execution.tasks.map((task: ExecutionTask) => {\n const newTask = { ...task };\n if (task.uiContext?.tree) {\n newTask.uiContext = {\n ...task.uiContext,\n tree: filterVisibleTree(task.uiContext.tree) || {\n node: null,\n children: [],\n },\n };\n }\n return newTask;\n })\n : execution.tasks,\n };\n}\n\ndeclare const __VERSION__: string | undefined;\n\nexport const getMidsceneVersion = (): string => {\n if (typeof __VERSION__ !== 'undefined') {\n return __VERSION__;\n } else if (\n process.env.__VERSION__ &&\n process.env.__VERSION__ !== 'undefined'\n ) {\n return process.env.__VERSION__;\n }\n throw new Error('__VERSION__ inject failed during build');\n};\n\nexport const parsePrompt = (\n prompt: TUserPrompt,\n): {\n textPrompt: string;\n multimodalPrompt?: TMultimodalPrompt;\n} => {\n if (typeof prompt === 'string') {\n return {\n textPrompt: prompt,\n multimodalPrompt: undefined,\n };\n }\n return {\n textPrompt: prompt.prompt,\n multimodalPrompt: prompt.images\n ? {\n images: prompt.images,\n convertHttpImage2Base64: !!prompt.convertHttpImage2Base64,\n }\n : undefined,\n };\n};\n"],"names":["__webpack_require__","module","getter","definition","key","Object","obj","prop","Symbol","debugProfile","getDebug","commonContextParser","interfaceInstance","assert","description","uploadTestInfoToServer","screenshotBase64","size","resizeImgBase64","getReportFileName","tag","reportTagName","globalConfigManager","MIDSCENE_REPORT_TAG_NAME","dateTimeInFileName","dayjs","uniqueId","uuid","printReportMsg","filepath","logMsg","getCurrentExecutionFile","trace","error","Error","stackTrace","pkgDir","process","stackLines","line","match","targetFileName","testFileIndex","Map","generateCacheId","fileName","taskFile","console","currentIndex","undefined","matchElementFromPlan","planLocateParam","tree","getNodeFromCacheList","centerPosition","Math","element","elementByPositionWithElementInfo","generateElementByPosition","matchElementFromCache","taskExecutor","xpaths","cachePrompt","cacheable","_taskExecutor_taskCache","i","cacheDebug","trimContextByViewport","execution","filterVisibleTree","node","filteredChildren","Array","child","task","_task_uiContext","newTask","getMidsceneVersion","__VERSION__","parsePrompt","prompt"],"mappings":";;;IACAA,oBAAoB,CAAC,GAAG,CAACC;QACxB,IAAIC,SAASD,UAAUA,OAAO,UAAU,GACvC,IAAOA,MAAM,CAAC,UAAU,GACxB,IAAOA;QACRD,oBAAoB,CAAC,CAACE,QAAQ;YAAE,GAAGA;QAAO;QAC1C,OAAOA;IACR;;;ICPAF,oBAAoB,CAAC,GAAG,CAAC,UAASG;QACjC,IAAI,IAAIC,OAAOD,WACR,IAAGH,oBAAoB,CAAC,CAACG,YAAYC,QAAQ,CAACJ,oBAAoB,CAAC,CAAC,UAASI,MACzEC,OAAO,cAAc,CAAC,UAASD,KAAK;YAAE,YAAY;YAAM,KAAKD,UAAU,CAACC,IAAI;QAAC;IAGzF;;;ICNAJ,oBAAoB,CAAC,GAAG,CAACM,KAAKC,OAAUF,OAAO,SAAS,CAAC,cAAc,CAAC,IAAI,CAACC,KAAKC;;;ICClFP,oBAAoB,CAAC,GAAG,CAAC;QACxB,IAAG,AAAkB,eAAlB,OAAOQ,UAA0BA,OAAO,WAAW,EACrDH,OAAO,cAAc,CAAC,UAASG,OAAO,WAAW,EAAE;YAAE,OAAO;QAAS;QAEtEH,OAAO,cAAc,CAAC,UAAS,cAAc;YAAE,OAAO;QAAK;IAC5D;;;;;;;;;;;;;;;;;;;;;;;;;;ACwBA,MAAMI,eAAeC,AAAAA,IAAAA,uBAAAA,QAAAA,AAAAA,EAAS;AAEvB,eAAeC,oBACpBC,iBAAoC;QAKhBA;IAHpBC,IAAAA,sBAAAA,MAAAA,AAAAA,EAAOD,mBAAmB;IAE1BH,aAAa;IACb,MAAMK,cAAcF,AAAAA,SAAAA,CAAAA,8BAAAA,kBAAkB,QAAQ,AAAD,IAAzBA,KAAAA,IAAAA,4BAAAA,IAAAA,CAAAA,kBAAiB,KAAiB;IACtDH,aAAa;IAEbA,aAAa;IACbM,IAAAA,kCAAAA,sBAAAA,AAAAA,EAAuB;QAAE,SAASD;IAAY;IAC9CL,aAAa;IAEb,IAAIO,mBAAmB,MAAMJ,kBAAkB,gBAAgB;IAC/DC,IAAAA,sBAAAA,MAAAA,AAAAA,EAAOG,kBAAmB;IAE1B,MAAMC,OAAO,MAAML,kBAAkB,IAAI;IACzCH,aAAa,CAAC,MAAM,EAAEQ,KAAK,KAAK,CAAC,CAAC,EAAEA,KAAK,MAAM,CAAC,MAAM,EAAEA,KAAK,GAAG,EAAE;IAElE,IAAIA,KAAK,GAAG,IAAIA,KAAK,GAAG,GAAG,GAAG;QAC5BR,aAAa;QACbO,mBAAmB,MAAME,AAAAA,IAAAA,oBAAAA,eAAAA,AAAAA,EAAgBF,kBAAkB;YACzD,OAAOC,KAAK,KAAK;YACjB,QAAQA,KAAK,MAAM;QACrB;QACAR,aAAa;IACf;IAEA,OAAO;QACL,MAAM;YACJ,MAAM;YACN,UAAU,EAAE;QACd;QACAQ;QACA,kBAAkBD;IACpB;AACF;AAEO,SAASG,kBAAkBC,MAAM,KAAK;IAC3C,MAAMC,gBAAgBC,oBAAAA,mBAAAA,CAAAA,iBAAqC,CACzDC,oBAAAA,wBAAwBA;IAE1B,MAAMC,qBAAqBC,2BAAQ,MAAM,CAAC;IAE1C,MAAMC,WAAWC,AAAAA,IAAAA,sBAAAA,IAAAA,AAAAA,IAAO,SAAS,CAAC,GAAG;IACrC,OAAO,GAAGN,iBAAiBD,IAAI,CAAC,EAAEI,mBAAmB,CAAC,EAAEE,UAAU;AACpE;AAEO,SAASE,eAAeC,QAAgB;IAC7CC,IAAAA,sBAAAA,MAAAA,AAAAA,EAAO,CAAC,gCAAgC,EAAED,UAAU;AACtD;AAMO,SAASE,wBAAwBC,KAAc;IACpD,MAAMC,QAAQ,IAAIC;IAClB,MAAMC,aAAaH,SAASC,MAAM,KAAK;IACvC,MAAMG,SAASC,QAAQ,GAAG,MAAM;IAChC,IAAIF,YAAY;QACd,MAAMG,aAAaH,WAAW,KAAK,CAAC;QACpC,KAAK,MAAMI,QAAQD,WACjB,IACEC,KAAK,QAAQ,CAAC,aACdA,KAAK,QAAQ,CAAC,aACdA,KAAK,QAAQ,CAAC,UACdA,KAAK,QAAQ,CAAC,QACd;YACA,MAAMC,QAAQD,KAAK,KAAK,CAAC;YACzB,IAAIC,QAAAA,QAAAA,KAAAA,IAAAA,KAAO,CAAC,EAAE,EAAE;gBACd,MAAMC,iBAAiBD,KAAK,CAAC,EAAE,CAC5B,OAAO,CAACJ,QAAQ,IAChB,IAAI,GACJ,OAAO,CAAC,OAAO;gBAClB,OAAOK;YACT;QACF;IAEJ;IACA,OAAO;AACT;AAEA,MAAMC,gBAAgB,IAAIC;AAEnB,SAASC,gBAAgBC,QAAiB;IAC/C,IAAIC,WAAWD,YAAYd;IAC3B,IAAI,CAACe,UAAU;QACbA,WAAWnB,AAAAA,IAAAA,sBAAAA,IAAAA,AAAAA;QACXoB,QAAQ,IAAI,CACV;IAEJ;IAEA,IAAIL,cAAc,GAAG,CAACI,WAAW;QAC/B,MAAME,eAAeN,cAAc,GAAG,CAACI;QACvC,IAAIE,AAAiBC,WAAjBD,cACFN,cAAc,GAAG,CAACI,UAAUE,eAAe;IAE/C,OACEN,cAAc,GAAG,CAACI,UAAU;IAE9B,OAAO,GAAGA,SAAS,CAAC,EAAEJ,cAAc,GAAG,CAACI,WAAW;AACrD;AAEO,SAASI,qBACdC,eAAoC,EACpCC,IAAkC;IAElC,IAAI,CAACD,iBACH;IAEF,IAAIA,gBAAgB,EAAE,EACpB,OAAOE,AAAAA,IAAAA,0BAAAA,oBAAAA,AAAAA,EAAqBF,gBAAgB,EAAE;IAGhD,IAAIA,gBAAgB,IAAI,EAAE;QACxB,MAAMG,iBAAiB;YACrB,GAAGC,KAAK,KAAK,CAAEJ,AAAAA,CAAAA,gBAAgB,IAAI,CAAC,EAAE,GAAGA,gBAAgB,IAAI,CAAC,EAAC,IAAK;YACpE,GAAGI,KAAK,KAAK,CAAEJ,AAAAA,CAAAA,gBAAgB,IAAI,CAAC,EAAE,GAAGA,gBAAgB,IAAI,CAAC,EAAC,IAAK;QACtE;QACA,IAAIK,UAAUC,AAAAA,IAAAA,yBAAAA,gCAAAA,AAAAA,EAAiCL,MAAME;QAErD,IAAI,CAACE,SACHA,UAAUE,AAAAA,IAAAA,0BAAAA,yBAAAA,AAAAA,EAA0BJ;QAGtC,OAAOE;IACT;AAGF;AAEO,eAAeG,sBACpBC,YAA0B,EAC1BC,MAA4B,EAC5BC,WAAwB,EACxBC,SAA8B;IAE9B,IAAI;YAGAC;QAFF,IACEH,AAAAA,CAAAA,QAAAA,SAAAA,KAAAA,IAAAA,OAAQ,MAAM,AAAD,KAAC,SACdG,CAAAA,0BAAAA,aAAa,SAAS,AAAD,IAArBA,KAAAA,IAAAA,wBAAwB,iBAAiB,AAAD,KACxCD,AAAc,UAAdA,aACCH,aAAa,SAAS,CAAS,qBAAqB,EAGrD,IAAK,IAAIK,IAAI,GAAGA,IAAIJ,OAAO,MAAM,EAAEI,IAAK;YACtC,MAAMT,UAAU,MACdI,aAAa,SAAS,CACtB,qBAAqB,CAACC,MAAM,CAACI,EAAE;YAEjC,IAAIT,QAAAA,UAAAA,KAAAA,IAAAA,QAAS,EAAE,EAAE;gBACfU,IAAAA,uCAAAA,KAAAA,AAAAA,EAAW,yBAAyBJ;gBACpCI,IAAAA,uCAAAA,KAAAA,AAAAA,EACE,0DACAL,MAAM,CAACI,EAAE,EACTT,QAAAA,UAAAA,KAAAA,IAAAA,QAAS,EAAE;gBAEb,OAAOA;YACT;QACF;IAEJ,EAAE,OAAOvB,OAAO;QACdiC,IAAAA,uCAAAA,KAAAA,AAAAA,EAAW,qCAAqCjC;IAClD;AACF;AAEO,SAASkC,sBAAsBC,SAAwB;IAC5D,SAASC,kBACPC,IAAkC;QAElC,IAAI,CAACA,MAAM,OAAO;QAGlB,MAAMC,mBAAmBC,MAAM,OAAO,CAACF,KAAK,QAAQ,IAC/CA,KAAK,QAAQ,CACX,GAAG,CAACD,mBACJ,MAAM,CAAC,CAACI,QAAUA,AAAU,SAAVA,SACrB,EAAE;QAGN,IAAIH,KAAK,IAAI,IAAIA,AAAwB,SAAxBA,KAAK,IAAI,CAAC,SAAS,EAClC,OAAO;YACL,GAAGA,IAAI;YACP,UAAUC;QACZ;QAIF,IAAIA,iBAAiB,MAAM,GAAG,GAC5B,OAAO;YACL,MAAM;YACN,UAAUA;QACZ;QAIF,OAAO;IACT;IAEA,OAAO;QACL,GAAGH,SAAS;QACZ,OAAOI,MAAM,OAAO,CAACJ,UAAU,KAAK,IAChCA,UAAU,KAAK,CAAC,GAAG,CAAC,CAACM;gBAEfC;YADJ,MAAMC,UAAU;gBAAE,GAAGF,IAAI;YAAC;YAC1B,IAAI,QAAAC,CAAAA,kBAAAA,KAAK,SAAS,AAAD,IAAbA,KAAAA,IAAAA,gBAAgB,IAAI,EACtBC,QAAQ,SAAS,GAAG;gBAClB,GAAGF,KAAK,SAAS;gBACjB,MAAML,kBAAkBK,KAAK,SAAS,CAAC,IAAI,KAAK;oBAC9C,MAAM;oBACN,UAAU,EAAE;gBACd;YACF;YAEF,OAAOE;QACT,KACAR,UAAU,KAAK;IACrB;AACF;AAIO,MAAMS,qBAAqB,IAEvBC;AAUJ,MAAMC,cAAc,CACzBC;IAKA,IAAI,AAAkB,YAAlB,OAAOA,QACT,OAAO;QACL,YAAYA;QACZ,kBAAkB/B;IACpB;IAEF,OAAO;QACL,YAAY+B,OAAO,MAAM;QACzB,kBAAkBA,OAAO,MAAM,GAC3B;YACE,QAAQA,OAAO,MAAM;YACrB,yBAAyB,CAAC,CAACA,OAAO,uBAAuB;QAC3D,IACA/B;IACN;AACF"}
1
+ {"version":3,"file":"agent/utils.js","sources":["webpack://@midscene/core/webpack/runtime/compat_get_default_export","webpack://@midscene/core/webpack/runtime/define_property_getters","webpack://@midscene/core/webpack/runtime/has_own_property","webpack://@midscene/core/webpack/runtime/make_namespace_object","webpack://@midscene/core/./src/agent/utils.ts"],"sourcesContent":["// getDefaultExport function for compatibility with non-ESM modules\n__webpack_require__.n = (module) => {\n\tvar getter = module && module.__esModule ?\n\t\t() => (module['default']) :\n\t\t() => (module);\n\t__webpack_require__.d(getter, { a: getter });\n\treturn getter;\n};\n","__webpack_require__.d = (exports, definition) => {\n\tfor(var key in definition) {\n if(__webpack_require__.o(definition, key) && !__webpack_require__.o(exports, key)) {\n Object.defineProperty(exports, key, { enumerable: true, get: definition[key] });\n }\n }\n};","__webpack_require__.o = (obj, prop) => (Object.prototype.hasOwnProperty.call(obj, prop))","// define __esModule on exports\n__webpack_require__.r = (exports) => {\n\tif(typeof Symbol !== 'undefined' && Symbol.toStringTag) {\n\t\tObject.defineProperty(exports, Symbol.toStringTag, { value: 'Module' });\n\t}\n\tObject.defineProperty(exports, '__esModule', { value: true });\n};","import { elementByPositionWithElementInfo } from '@/ai-model';\nimport type { AbstractInterface } from '@/device';\nimport type {\n BaseElement,\n ElementTreeNode,\n ExecutionDump,\n ExecutionTask,\n ExecutorContext,\n PlanningLocateParam,\n TMultimodalPrompt,\n TUserPrompt,\n UIContext,\n} from '@/index';\nimport { uploadTestInfoToServer } from '@/utils';\nimport {\n MIDSCENE_REPORT_TAG_NAME,\n globalConfigManager,\n} from '@midscene/shared/env';\nimport {\n generateElementByPosition,\n getNodeFromCacheList,\n} from '@midscene/shared/extractor';\nimport { resizeImgBase64 } from '@midscene/shared/img';\nimport { getDebug } from '@midscene/shared/logger';\nimport { _keyDefinitions } from '@midscene/shared/us-keyboard-layout';\nimport { assert, logMsg, uuid } from '@midscene/shared/utils';\nimport dayjs from 'dayjs';\nimport { debug as cacheDebug } from './task-cache';\nimport type { TaskExecutor } from './tasks';\n\nconst debugProfile = getDebug('web:tool:profile');\n\nexport async function commonContextParser(\n interfaceInstance: AbstractInterface,\n _opt: { uploadServerUrl?: string },\n): Promise<UIContext> {\n assert(interfaceInstance, 'interfaceInstance is required');\n\n debugProfile('Getting interface description');\n const description = interfaceInstance.describe?.() || '';\n debugProfile('Interface description end');\n\n debugProfile('Uploading test info to server');\n uploadTestInfoToServer({\n testUrl: description,\n serverUrl: _opt.uploadServerUrl,\n });\n debugProfile('UploadTestInfoToServer end');\n\n let screenshotBase64 = await interfaceInstance.screenshotBase64();\n assert(screenshotBase64!, 'screenshotBase64 is required');\n\n const size = await interfaceInstance.size();\n debugProfile(`size: ${size.width}x${size.height} dpr: ${size.dpr}`);\n\n if (size.dpr && size.dpr > 1) {\n debugProfile('Resizing screenshot for high DPR display');\n screenshotBase64 = await resizeImgBase64(screenshotBase64, {\n width: size.width,\n height: size.height,\n });\n debugProfile('ResizeImgBase64 end');\n }\n\n return {\n tree: {\n node: null,\n children: [],\n },\n size,\n screenshotBase64: screenshotBase64!,\n };\n}\n\nexport function getReportFileName(tag = 'web') {\n const reportTagName = globalConfigManager.getEnvConfigValue(\n MIDSCENE_REPORT_TAG_NAME,\n );\n const dateTimeInFileName = dayjs().format('YYYY-MM-DD_HH-mm-ss');\n // ensure uniqueness at the same time\n const uniqueId = uuid().substring(0, 8);\n return `${reportTagName || tag}-${dateTimeInFileName}-${uniqueId}`;\n}\n\nexport function printReportMsg(filepath: string) {\n logMsg(`Midscene - report file updated: ${filepath}`);\n}\n\n/**\n * Get the current execution file name\n * @returns The name of the current execution file\n */\nexport function getCurrentExecutionFile(trace?: string): string | false {\n const error = new Error();\n const stackTrace = trace || error.stack;\n const pkgDir = process.cwd() || '';\n if (stackTrace) {\n const stackLines = stackTrace.split('\\n');\n for (const line of stackLines) {\n if (\n line.includes('.spec.') ||\n line.includes('.test.') ||\n line.includes('.ts') ||\n line.includes('.js')\n ) {\n const match = line.match(/(?:at\\s+)?(.*?\\.(?:spec|test)\\.[jt]s)/);\n if (match?.[1]) {\n const targetFileName = match[1]\n .replace(pkgDir, '')\n .trim()\n .replace('at ', '');\n return targetFileName;\n }\n }\n }\n }\n return false;\n}\n\nconst testFileIndex = new Map<string, number>();\n\nexport function generateCacheId(fileName?: string): string {\n let taskFile = fileName || getCurrentExecutionFile();\n if (!taskFile) {\n taskFile = uuid();\n console.warn(\n 'Midscene - using random UUID for cache id. Cache may be invalid.',\n );\n }\n\n if (testFileIndex.has(taskFile)) {\n const currentIndex = testFileIndex.get(taskFile);\n if (currentIndex !== undefined) {\n testFileIndex.set(taskFile, currentIndex + 1);\n }\n } else {\n testFileIndex.set(taskFile, 1);\n }\n return `${taskFile}-${testFileIndex.get(taskFile)}`;\n}\n\nexport function matchElementFromPlan(\n planLocateParam: PlanningLocateParam,\n tree: ElementTreeNode<BaseElement>,\n) {\n if (!planLocateParam) {\n return undefined;\n }\n if (planLocateParam.id) {\n return getNodeFromCacheList(planLocateParam.id);\n }\n\n if (planLocateParam.bbox) {\n const centerPosition = {\n x: Math.floor((planLocateParam.bbox[0] + planLocateParam.bbox[2]) / 2),\n y: Math.floor((planLocateParam.bbox[1] + planLocateParam.bbox[3]) / 2),\n };\n let element = elementByPositionWithElementInfo(tree, centerPosition);\n\n if (!element) {\n element = generateElementByPosition(centerPosition) as BaseElement;\n }\n\n return element;\n }\n\n return undefined;\n}\n\nexport async function matchElementFromCache(\n taskExecutor: TaskExecutor,\n xpaths: string[] | undefined,\n cachePrompt: TUserPrompt,\n cacheable: boolean | undefined,\n) {\n try {\n if (\n xpaths?.length &&\n taskExecutor.taskCache?.isCacheResultUsed &&\n cacheable !== false &&\n (taskExecutor.interface as any).getElementInfoByXpath\n ) {\n // hit cache, use new id\n for (let i = 0; i < xpaths.length; i++) {\n const element = await (\n taskExecutor.interface as any\n ).getElementInfoByXpath(xpaths[i]);\n\n if (element?.id) {\n cacheDebug('cache hit, prompt: %s', cachePrompt);\n cacheDebug(\n 'found a new element with same xpath, xpath: %s, id: %s',\n xpaths[i],\n element?.id,\n );\n return element;\n }\n }\n }\n } catch (error) {\n cacheDebug('get element info by xpath error: ', error);\n }\n}\n\nexport function trimContextByViewport(execution: ExecutionDump) {\n function filterVisibleTree(\n node: ElementTreeNode<BaseElement>,\n ): ElementTreeNode<BaseElement> | null {\n if (!node) return null;\n\n // recursively process all children\n const filteredChildren = Array.isArray(node.children)\n ? (node.children\n .map(filterVisibleTree)\n .filter((child) => child !== null) as ElementTreeNode<BaseElement>[])\n : [];\n\n // if the current node is visible, keep it and the filtered children\n if (node.node && node.node.isVisible === true) {\n return {\n ...node,\n children: filteredChildren,\n };\n }\n\n // if the current node is invisible, but has visible children, create an empty node to include these children\n if (filteredChildren.length > 0) {\n return {\n node: null,\n children: filteredChildren,\n };\n }\n\n // if the current node is invisible and has no visible children, return null\n return null;\n }\n\n return {\n ...execution,\n tasks: Array.isArray(execution.tasks)\n ? execution.tasks.map((task: ExecutionTask) => {\n const newTask = { ...task };\n if (task.uiContext?.tree) {\n newTask.uiContext = {\n ...task.uiContext,\n tree: filterVisibleTree(task.uiContext.tree) || {\n node: null,\n children: [],\n },\n };\n }\n return newTask;\n })\n : execution.tasks,\n };\n}\n\ndeclare const __VERSION__: string | undefined;\n\nexport const getMidsceneVersion = (): string => {\n if (typeof __VERSION__ !== 'undefined') {\n return __VERSION__;\n } else if (\n process.env.__VERSION__ &&\n process.env.__VERSION__ !== 'undefined'\n ) {\n return process.env.__VERSION__;\n }\n throw new Error('__VERSION__ inject failed during build');\n};\n\nexport const parsePrompt = (\n prompt: TUserPrompt,\n): {\n textPrompt: string;\n multimodalPrompt?: TMultimodalPrompt;\n} => {\n if (typeof prompt === 'string') {\n return {\n textPrompt: prompt,\n multimodalPrompt: undefined,\n };\n }\n return {\n textPrompt: prompt.prompt,\n multimodalPrompt: prompt.images\n ? {\n images: prompt.images,\n convertHttpImage2Base64: !!prompt.convertHttpImage2Base64,\n }\n : undefined,\n };\n};\n"],"names":["__webpack_require__","module","getter","definition","key","Object","obj","prop","Symbol","debugProfile","getDebug","commonContextParser","interfaceInstance","_opt","assert","description","uploadTestInfoToServer","screenshotBase64","size","resizeImgBase64","getReportFileName","tag","reportTagName","globalConfigManager","MIDSCENE_REPORT_TAG_NAME","dateTimeInFileName","dayjs","uniqueId","uuid","printReportMsg","filepath","logMsg","getCurrentExecutionFile","trace","error","Error","stackTrace","pkgDir","process","stackLines","line","match","targetFileName","testFileIndex","Map","generateCacheId","fileName","taskFile","console","currentIndex","undefined","matchElementFromPlan","planLocateParam","tree","getNodeFromCacheList","centerPosition","Math","element","elementByPositionWithElementInfo","generateElementByPosition","matchElementFromCache","taskExecutor","xpaths","cachePrompt","cacheable","_taskExecutor_taskCache","i","cacheDebug","trimContextByViewport","execution","filterVisibleTree","node","filteredChildren","Array","child","task","_task_uiContext","newTask","getMidsceneVersion","__VERSION__","parsePrompt","prompt"],"mappings":";;;IACAA,oBAAoB,CAAC,GAAG,CAACC;QACxB,IAAIC,SAASD,UAAUA,OAAO,UAAU,GACvC,IAAOA,MAAM,CAAC,UAAU,GACxB,IAAOA;QACRD,oBAAoB,CAAC,CAACE,QAAQ;YAAE,GAAGA;QAAO;QAC1C,OAAOA;IACR;;;ICPAF,oBAAoB,CAAC,GAAG,CAAC,UAASG;QACjC,IAAI,IAAIC,OAAOD,WACR,IAAGH,oBAAoB,CAAC,CAACG,YAAYC,QAAQ,CAACJ,oBAAoB,CAAC,CAAC,UAASI,MACzEC,OAAO,cAAc,CAAC,UAASD,KAAK;YAAE,YAAY;YAAM,KAAKD,UAAU,CAACC,IAAI;QAAC;IAGzF;;;ICNAJ,oBAAoB,CAAC,GAAG,CAACM,KAAKC,OAAUF,OAAO,SAAS,CAAC,cAAc,CAAC,IAAI,CAACC,KAAKC;;;ICClFP,oBAAoB,CAAC,GAAG,CAAC;QACxB,IAAG,AAAkB,eAAlB,OAAOQ,UAA0BA,OAAO,WAAW,EACrDH,OAAO,cAAc,CAAC,UAASG,OAAO,WAAW,EAAE;YAAE,OAAO;QAAS;QAEtEH,OAAO,cAAc,CAAC,UAAS,cAAc;YAAE,OAAO;QAAK;IAC5D;;;;;;;;;;;;;;;;;;;;;;;;;;ACwBA,MAAMI,eAAeC,AAAAA,IAAAA,uBAAAA,QAAAA,AAAAA,EAAS;AAEvB,eAAeC,oBACpBC,iBAAoC,EACpCC,IAAkC;QAKdD;IAHpBE,IAAAA,sBAAAA,MAAAA,AAAAA,EAAOF,mBAAmB;IAE1BH,aAAa;IACb,MAAMM,cAAcH,AAAAA,SAAAA,CAAAA,8BAAAA,kBAAkB,QAAQ,AAAD,IAAzBA,KAAAA,IAAAA,4BAAAA,IAAAA,CAAAA,kBAAiB,KAAiB;IACtDH,aAAa;IAEbA,aAAa;IACbO,IAAAA,kCAAAA,sBAAAA,AAAAA,EAAuB;QACrB,SAASD;QACT,WAAWF,KAAK,eAAe;IACjC;IACAJ,aAAa;IAEb,IAAIQ,mBAAmB,MAAML,kBAAkB,gBAAgB;IAC/DE,IAAAA,sBAAAA,MAAAA,AAAAA,EAAOG,kBAAmB;IAE1B,MAAMC,OAAO,MAAMN,kBAAkB,IAAI;IACzCH,aAAa,CAAC,MAAM,EAAES,KAAK,KAAK,CAAC,CAAC,EAAEA,KAAK,MAAM,CAAC,MAAM,EAAEA,KAAK,GAAG,EAAE;IAElE,IAAIA,KAAK,GAAG,IAAIA,KAAK,GAAG,GAAG,GAAG;QAC5BT,aAAa;QACbQ,mBAAmB,MAAME,AAAAA,IAAAA,oBAAAA,eAAAA,AAAAA,EAAgBF,kBAAkB;YACzD,OAAOC,KAAK,KAAK;YACjB,QAAQA,KAAK,MAAM;QACrB;QACAT,aAAa;IACf;IAEA,OAAO;QACL,MAAM;YACJ,MAAM;YACN,UAAU,EAAE;QACd;QACAS;QACA,kBAAkBD;IACpB;AACF;AAEO,SAASG,kBAAkBC,MAAM,KAAK;IAC3C,MAAMC,gBAAgBC,oBAAAA,mBAAAA,CAAAA,iBAAqC,CACzDC,oBAAAA,wBAAwBA;IAE1B,MAAMC,qBAAqBC,2BAAQ,MAAM,CAAC;IAE1C,MAAMC,WAAWC,AAAAA,IAAAA,sBAAAA,IAAAA,AAAAA,IAAO,SAAS,CAAC,GAAG;IACrC,OAAO,GAAGN,iBAAiBD,IAAI,CAAC,EAAEI,mBAAmB,CAAC,EAAEE,UAAU;AACpE;AAEO,SAASE,eAAeC,QAAgB;IAC7CC,IAAAA,sBAAAA,MAAAA,AAAAA,EAAO,CAAC,gCAAgC,EAAED,UAAU;AACtD;AAMO,SAASE,wBAAwBC,KAAc;IACpD,MAAMC,QAAQ,IAAIC;IAClB,MAAMC,aAAaH,SAASC,MAAM,KAAK;IACvC,MAAMG,SAASC,QAAQ,GAAG,MAAM;IAChC,IAAIF,YAAY;QACd,MAAMG,aAAaH,WAAW,KAAK,CAAC;QACpC,KAAK,MAAMI,QAAQD,WACjB,IACEC,KAAK,QAAQ,CAAC,aACdA,KAAK,QAAQ,CAAC,aACdA,KAAK,QAAQ,CAAC,UACdA,KAAK,QAAQ,CAAC,QACd;YACA,MAAMC,QAAQD,KAAK,KAAK,CAAC;YACzB,IAAIC,QAAAA,QAAAA,KAAAA,IAAAA,KAAO,CAAC,EAAE,EAAE;gBACd,MAAMC,iBAAiBD,KAAK,CAAC,EAAE,CAC5B,OAAO,CAACJ,QAAQ,IAChB,IAAI,GACJ,OAAO,CAAC,OAAO;gBAClB,OAAOK;YACT;QACF;IAEJ;IACA,OAAO;AACT;AAEA,MAAMC,gBAAgB,IAAIC;AAEnB,SAASC,gBAAgBC,QAAiB;IAC/C,IAAIC,WAAWD,YAAYd;IAC3B,IAAI,CAACe,UAAU;QACbA,WAAWnB,AAAAA,IAAAA,sBAAAA,IAAAA,AAAAA;QACXoB,QAAQ,IAAI,CACV;IAEJ;IAEA,IAAIL,cAAc,GAAG,CAACI,WAAW;QAC/B,MAAME,eAAeN,cAAc,GAAG,CAACI;QACvC,IAAIE,AAAiBC,WAAjBD,cACFN,cAAc,GAAG,CAACI,UAAUE,eAAe;IAE/C,OACEN,cAAc,GAAG,CAACI,UAAU;IAE9B,OAAO,GAAGA,SAAS,CAAC,EAAEJ,cAAc,GAAG,CAACI,WAAW;AACrD;AAEO,SAASI,qBACdC,eAAoC,EACpCC,IAAkC;IAElC,IAAI,CAACD,iBACH;IAEF,IAAIA,gBAAgB,EAAE,EACpB,OAAOE,AAAAA,IAAAA,0BAAAA,oBAAAA,AAAAA,EAAqBF,gBAAgB,EAAE;IAGhD,IAAIA,gBAAgB,IAAI,EAAE;QACxB,MAAMG,iBAAiB;YACrB,GAAGC,KAAK,KAAK,CAAEJ,AAAAA,CAAAA,gBAAgB,IAAI,CAAC,EAAE,GAAGA,gBAAgB,IAAI,CAAC,EAAC,IAAK;YACpE,GAAGI,KAAK,KAAK,CAAEJ,AAAAA,CAAAA,gBAAgB,IAAI,CAAC,EAAE,GAAGA,gBAAgB,IAAI,CAAC,EAAC,IAAK;QACtE;QACA,IAAIK,UAAUC,AAAAA,IAAAA,yBAAAA,gCAAAA,AAAAA,EAAiCL,MAAME;QAErD,IAAI,CAACE,SACHA,UAAUE,AAAAA,IAAAA,0BAAAA,yBAAAA,AAAAA,EAA0BJ;QAGtC,OAAOE;IACT;AAGF;AAEO,eAAeG,sBACpBC,YAA0B,EAC1BC,MAA4B,EAC5BC,WAAwB,EACxBC,SAA8B;IAE9B,IAAI;YAGAC;QAFF,IACEH,AAAAA,CAAAA,QAAAA,SAAAA,KAAAA,IAAAA,OAAQ,MAAM,AAAD,KAAC,SACdG,CAAAA,0BAAAA,aAAa,SAAS,AAAD,IAArBA,KAAAA,IAAAA,wBAAwB,iBAAiB,AAAD,KACxCD,AAAc,UAAdA,aACCH,aAAa,SAAS,CAAS,qBAAqB,EAGrD,IAAK,IAAIK,IAAI,GAAGA,IAAIJ,OAAO,MAAM,EAAEI,IAAK;YACtC,MAAMT,UAAU,MACdI,aAAa,SAAS,CACtB,qBAAqB,CAACC,MAAM,CAACI,EAAE;YAEjC,IAAIT,QAAAA,UAAAA,KAAAA,IAAAA,QAAS,EAAE,EAAE;gBACfU,IAAAA,uCAAAA,KAAAA,AAAAA,EAAW,yBAAyBJ;gBACpCI,IAAAA,uCAAAA,KAAAA,AAAAA,EACE,0DACAL,MAAM,CAACI,EAAE,EACTT,QAAAA,UAAAA,KAAAA,IAAAA,QAAS,EAAE;gBAEb,OAAOA;YACT;QACF;IAEJ,EAAE,OAAOvB,OAAO;QACdiC,IAAAA,uCAAAA,KAAAA,AAAAA,EAAW,qCAAqCjC;IAClD;AACF;AAEO,SAASkC,sBAAsBC,SAAwB;IAC5D,SAASC,kBACPC,IAAkC;QAElC,IAAI,CAACA,MAAM,OAAO;QAGlB,MAAMC,mBAAmBC,MAAM,OAAO,CAACF,KAAK,QAAQ,IAC/CA,KAAK,QAAQ,CACX,GAAG,CAACD,mBACJ,MAAM,CAAC,CAACI,QAAUA,AAAU,SAAVA,SACrB,EAAE;QAGN,IAAIH,KAAK,IAAI,IAAIA,AAAwB,SAAxBA,KAAK,IAAI,CAAC,SAAS,EAClC,OAAO;YACL,GAAGA,IAAI;YACP,UAAUC;QACZ;QAIF,IAAIA,iBAAiB,MAAM,GAAG,GAC5B,OAAO;YACL,MAAM;YACN,UAAUA;QACZ;QAIF,OAAO;IACT;IAEA,OAAO;QACL,GAAGH,SAAS;QACZ,OAAOI,MAAM,OAAO,CAACJ,UAAU,KAAK,IAChCA,UAAU,KAAK,CAAC,GAAG,CAAC,CAACM;gBAEfC;YADJ,MAAMC,UAAU;gBAAE,GAAGF,IAAI;YAAC;YAC1B,IAAI,QAAAC,CAAAA,kBAAAA,KAAK,SAAS,AAAD,IAAbA,KAAAA,IAAAA,gBAAgB,IAAI,EACtBC,QAAQ,SAAS,GAAG;gBAClB,GAAGF,KAAK,SAAS;gBACjB,MAAML,kBAAkBK,KAAK,SAAS,CAAC,IAAI,KAAK;oBAC9C,MAAM;oBACN,UAAU,EAAE;gBACd;YACF;YAEF,OAAOE;QACT,KACAR,UAAU,KAAK;IACrB;AACF;AAIO,MAAMS,qBAAqB,IAEvBC;AAUJ,MAAMC,cAAc,CACzBC;IAKA,IAAI,AAAkB,YAAlB,OAAOA,QACT,OAAO;QACL,YAAYA;QACZ,kBAAkB/B;IACpB;IAEF,OAAO;QACL,YAAY+B,OAAO,MAAM;QACzB,kBAAkBA,OAAO,MAAM,GAC3B;YACE,QAAQA,OAAO,MAAM;YACrB,yBAAyB,CAAC,CAACA,OAAO,uBAAuB;QAC3D,IACA/B;IACN;AACF"}
@@ -40,19 +40,16 @@ __webpack_require__.d(__webpack_exports__, {
40
40
  adaptBboxToRect: ()=>adaptBboxToRect,
41
41
  markupImageForLLM: ()=>markupImageForLLM,
42
42
  adaptDoubaoBbox: ()=>adaptDoubaoBbox,
43
- callAiFn: ()=>callAiFn,
44
43
  dumpMidsceneLocatorField: ()=>dumpMidsceneLocatorField,
44
+ getMidsceneLocationSchema: ()=>getMidsceneLocationSchema,
45
45
  adaptQwenBbox: ()=>adaptQwenBbox,
46
46
  buildYamlFlowFromPlans: ()=>buildYamlFlowFromPlans,
47
- getMidsceneLocationSchema: ()=>getMidsceneLocationSchema,
48
47
  ifMidsceneLocatorField: ()=>ifMidsceneLocatorField,
49
- dumpActionParam: ()=>dumpActionParam,
50
- warnGPT4oSizeLimit: ()=>warnGPT4oSizeLimit
48
+ warnGPT4oSizeLimit: ()=>warnGPT4oSizeLimit,
49
+ dumpActionParam: ()=>dumpActionParam
51
50
  });
52
51
  const utils_namespaceObject = require("@midscene/shared/utils");
53
- const index_js_namespaceObject = require("./service-caller/index.js");
54
52
  const constants_namespaceObject = require("@midscene/shared/constants");
55
- const env_namespaceObject = require("@midscene/shared/env");
56
53
  const extractor_namespaceObject = require("@midscene/shared/extractor");
57
54
  const img_namespaceObject = require("@midscene/shared/img");
58
55
  const logger_namespaceObject = require("@midscene/shared/logger");
@@ -65,21 +62,14 @@ var common_AIActionType = /*#__PURE__*/ function(AIActionType) {
65
62
  AIActionType[AIActionType["DESCRIBE_ELEMENT"] = 4] = "DESCRIBE_ELEMENT";
66
63
  return AIActionType;
67
64
  }({});
68
- async function callAiFn(msgs, AIActionTypeValue, modelPreferences) {
69
- const jsonObject = await (0, index_js_namespaceObject.callToGetJSONObject)(msgs, AIActionTypeValue, modelPreferences);
70
- return {
71
- content: jsonObject.content,
72
- usage: jsonObject.usage
73
- };
74
- }
75
65
  const defaultBboxSize = 20;
76
66
  const debugInspectUtils = (0, logger_namespaceObject.getDebug)('ai:common');
77
- function fillBboxParam(locate, width, height, modelPreferences) {
67
+ function fillBboxParam(locate, width, height, vlMode) {
78
68
  if (locate.bbox_2d && !(null == locate ? void 0 : locate.bbox)) {
79
69
  locate.bbox = locate.bbox_2d;
80
70
  delete locate.bbox_2d;
81
71
  }
82
- if (null == locate ? void 0 : locate.bbox) locate.bbox = adaptBbox(locate.bbox, width, height, modelPreferences);
72
+ if (null == locate ? void 0 : locate.bbox) locate.bbox = adaptBbox(locate.bbox, width, height, vlMode);
83
73
  return locate;
84
74
  }
85
75
  function adaptQwenBbox(bbox) {
@@ -141,9 +131,9 @@ function adaptDoubaoBbox(bbox, width, height) {
141
131
  const msg = `invalid bbox data for doubao-vision mode: ${JSON.stringify(bbox)} `;
142
132
  throw new Error(msg);
143
133
  }
144
- function adaptBbox(bbox, width, height, modelPreferences) {
145
- if ('doubao-vision' === (0, env_namespaceObject.vlLocateMode)(modelPreferences) || 'vlm-ui-tars' === (0, env_namespaceObject.vlLocateMode)(modelPreferences)) return adaptDoubaoBbox(bbox, width, height);
146
- if ('gemini' === (0, env_namespaceObject.vlLocateMode)(modelPreferences)) return adaptGeminiBbox(bbox, width, height);
134
+ function adaptBbox(bbox, width, height, vlMode) {
135
+ if ('doubao-vision' === vlMode || 'vlm-ui-tars' === vlMode) return adaptDoubaoBbox(bbox, width, height);
136
+ if ('gemini' === vlMode) return adaptGeminiBbox(bbox, width, height);
147
137
  return adaptQwenBbox(bbox);
148
138
  }
149
139
  function adaptGeminiBbox(bbox, width, height) {
@@ -158,9 +148,9 @@ function adaptGeminiBbox(bbox, width, height) {
158
148
  bottom
159
149
  ];
160
150
  }
161
- function adaptBboxToRect(bbox, width, height, modelPreferences, offsetX = 0, offsetY = 0) {
151
+ function adaptBboxToRect(bbox, width, height, offsetX = 0, offsetY = 0, vlMode) {
162
152
  debugInspectUtils('adaptBboxToRect', bbox, width, height, offsetX, offsetY);
163
- const [left, top, right, bottom] = adaptBbox(bbox, width, height, modelPreferences);
153
+ const [left, top, right, bottom] = adaptBbox(bbox, width, height, vlMode);
164
154
  const rectLeft = left;
165
155
  const rectTop = top;
166
156
  let rectWidth = right - left;
@@ -179,10 +169,9 @@ function adaptBboxToRect(bbox, width, height, modelPreferences, offsetX = 0, off
179
169
  return rect;
180
170
  }
181
171
  let warned = false;
182
- function warnGPT4oSizeLimit(size, modelPreferences) {
183
- var _getModelName;
172
+ function warnGPT4oSizeLimit(size, modelName) {
184
173
  if (warned) return;
185
- if (null == (_getModelName = (0, env_namespaceObject.getModelName)(modelPreferences)) ? void 0 : _getModelName.toLowerCase().includes('gpt-4o')) {
174
+ if (modelName.toLowerCase().includes('gpt-4o')) {
186
175
  const warningMsg = `GPT-4o has a maximum image input size of 2000x768 or 768x2000, but got ${size.width}x${size.height}. Please set your interface to a smaller resolution. Otherwise, the result may be inaccurate.`;
187
176
  if (Math.max(size.width, size.height) > 2000 || Math.min(size.width, size.height) > 768) {
188
177
  console.warn(warningMsg);
@@ -205,8 +194,8 @@ function mergeRects(rects) {
205
194
  height: maxBottom - minTop
206
195
  };
207
196
  }
208
- function expandSearchArea(rect, screenSize, modelPreferences) {
209
- const minEdgeSize = 'doubao-vision' === (0, env_namespaceObject.vlLocateMode)(modelPreferences) ? 500 : 300;
197
+ function expandSearchArea(rect, screenSize, vlMode) {
198
+ const minEdgeSize = 'doubao-vision' === vlMode ? 500 : 300;
210
199
  const defaultPadding = 160;
211
200
  const paddingSizeHorizontal = rect.width < minEdgeSize ? Math.ceil((minEdgeSize - rect.width) / 2) : defaultPadding;
212
201
  const paddingSizeVertical = rect.height < minEdgeSize ? Math.ceil((minEdgeSize - rect.height) / 2) : defaultPadding;
@@ -372,7 +361,6 @@ exports.adaptDoubaoBbox = __webpack_exports__.adaptDoubaoBbox;
372
361
  exports.adaptGeminiBbox = __webpack_exports__.adaptGeminiBbox;
373
362
  exports.adaptQwenBbox = __webpack_exports__.adaptQwenBbox;
374
363
  exports.buildYamlFlowFromPlans = __webpack_exports__.buildYamlFlowFromPlans;
375
- exports.callAiFn = __webpack_exports__.callAiFn;
376
364
  exports.dumpActionParam = __webpack_exports__.dumpActionParam;
377
365
  exports.dumpMidsceneLocatorField = __webpack_exports__.dumpMidsceneLocatorField;
378
366
  exports.expandSearchArea = __webpack_exports__.expandSearchArea;
@@ -397,7 +385,6 @@ for(var __webpack_i__ in __webpack_exports__)if (-1 === [
397
385
  "adaptGeminiBbox",
398
386
  "adaptQwenBbox",
399
387
  "buildYamlFlowFromPlans",
400
- "callAiFn",
401
388
  "dumpActionParam",
402
389
  "dumpMidsceneLocatorField",
403
390
  "expandSearchArea",
@@ -1 +1 @@
1
- {"version":3,"file":"ai-model/common.js","sources":["webpack://@midscene/core/webpack/runtime/define_property_getters","webpack://@midscene/core/webpack/runtime/has_own_property","webpack://@midscene/core/webpack/runtime/make_namespace_object","webpack://@midscene/core/./src/ai-model/common.ts"],"sourcesContent":["__webpack_require__.d = (exports, definition) => {\n\tfor(var key in definition) {\n if(__webpack_require__.o(definition, key) && !__webpack_require__.o(exports, key)) {\n Object.defineProperty(exports, key, { enumerable: true, get: definition[key] });\n }\n }\n};","__webpack_require__.o = (obj, prop) => (Object.prototype.hasOwnProperty.call(obj, prop))","// define __esModule on exports\n__webpack_require__.r = (exports) => {\n\tif(typeof Symbol !== 'undefined' && Symbol.toStringTag) {\n\t\tObject.defineProperty(exports, Symbol.toStringTag, { value: 'Module' });\n\t}\n\tObject.defineProperty(exports, '__esModule', { value: true });\n};","import type {\n AIUsageInfo,\n BaseElement,\n DeviceAction,\n ElementTreeNode,\n MidsceneYamlFlowItem,\n PlanningAction,\n Rect,\n Size,\n} from '@/types';\nimport { assert } from '@midscene/shared/utils';\n\nimport type {\n ChatCompletionSystemMessageParam,\n ChatCompletionUserMessageParam,\n} from 'openai/resources/index';\nimport { callToGetJSONObject } from './service-caller/index';\n\nimport type { PlanningLocateParam } from '@/types';\nimport { NodeType } from '@midscene/shared/constants';\nimport {\n type IModelPreferences,\n getModelName,\n vlLocateMode,\n} from '@midscene/shared/env';\nimport { treeToList } from '@midscene/shared/extractor';\nimport { compositeElementInfoImg } from '@midscene/shared/img';\nimport { getDebug } from '@midscene/shared/logger';\nimport { z } from 'zod';\n\nexport type AIArgs = [\n ChatCompletionSystemMessageParam,\n ...ChatCompletionUserMessageParam[],\n];\n\nexport enum AIActionType {\n ASSERT = 0,\n INSPECT_ELEMENT = 1,\n EXTRACT_DATA = 2,\n PLAN = 3,\n DESCRIBE_ELEMENT = 4,\n}\n\nexport async function callAiFn<T>(\n msgs: AIArgs,\n AIActionTypeValue: AIActionType,\n modelPreferences: IModelPreferences,\n): Promise<{ content: T; usage?: AIUsageInfo }> {\n const jsonObject = await callToGetJSONObject<T>(\n msgs,\n AIActionTypeValue,\n modelPreferences,\n );\n\n return {\n content: jsonObject.content,\n usage: jsonObject.usage,\n };\n}\n\nconst defaultBboxSize = 20; // must be even number\nconst debugInspectUtils = getDebug('ai:common');\n\n// transform the param of locate from qwen mode\nexport function fillBboxParam(\n locate: PlanningLocateParam,\n width: number,\n height: number,\n modelPreferences: IModelPreferences,\n) {\n // The Qwen model might have hallucinations of naming bbox as bbox_2d.\n if ((locate as any).bbox_2d && !locate?.bbox) {\n locate.bbox = (locate as any).bbox_2d;\n // biome-ignore lint/performance/noDelete: <explanation>\n delete (locate as any).bbox_2d;\n }\n\n if (locate?.bbox) {\n locate.bbox = adaptBbox(locate.bbox, width, height, modelPreferences);\n }\n\n return locate;\n}\n\nexport function adaptQwenBbox(\n bbox: number[],\n): [number, number, number, number] {\n if (bbox.length < 2) {\n const msg = `invalid bbox data for qwen-vl mode: ${JSON.stringify(bbox)} `;\n throw new Error(msg);\n }\n\n const result: [number, number, number, number] = [\n Math.round(bbox[0]),\n Math.round(bbox[1]),\n typeof bbox[2] === 'number'\n ? Math.round(bbox[2])\n : Math.round(bbox[0] + defaultBboxSize),\n typeof bbox[3] === 'number'\n ? Math.round(bbox[3])\n : Math.round(bbox[1] + defaultBboxSize),\n ];\n return result;\n}\n\nexport function adaptDoubaoBbox(\n bbox: string[] | number[] | string,\n width: number,\n height: number,\n): [number, number, number, number] {\n assert(\n width > 0 && height > 0,\n 'width and height must be greater than 0 in doubao mode',\n );\n\n if (typeof bbox === 'string') {\n assert(\n /^(\\d+)\\s(\\d+)\\s(\\d+)\\s(\\d+)$/.test(bbox.trim()),\n `invalid bbox data string for doubao-vision mode: ${bbox}`,\n );\n const splitted = bbox.split(' ');\n if (splitted.length === 4) {\n return [\n Math.round((Number(splitted[0]) * width) / 1000),\n Math.round((Number(splitted[1]) * height) / 1000),\n Math.round((Number(splitted[2]) * width) / 1000),\n Math.round((Number(splitted[3]) * height) / 1000),\n ];\n }\n throw new Error(`invalid bbox data string for doubao-vision mode: ${bbox}`);\n }\n\n if (Array.isArray(bbox) && Array.isArray(bbox[0])) {\n bbox = bbox[0];\n }\n\n let bboxList: number[] = [];\n if (Array.isArray(bbox) && typeof bbox[0] === 'string') {\n bbox.forEach((item) => {\n if (typeof item === 'string' && item.includes(',')) {\n const [x, y] = item.split(',');\n bboxList.push(Number(x.trim()), Number(y.trim()));\n } else if (typeof item === 'string' && item.includes(' ')) {\n const [x, y] = item.split(' ');\n bboxList.push(Number(x.trim()), Number(y.trim()));\n } else {\n bboxList.push(Number(item));\n }\n });\n } else {\n bboxList = bbox as any;\n }\n\n if (bboxList.length === 4 || bboxList.length === 5) {\n return [\n Math.round((bboxList[0] * width) / 1000),\n Math.round((bboxList[1] * height) / 1000),\n Math.round((bboxList[2] * width) / 1000),\n Math.round((bboxList[3] * height) / 1000),\n ];\n }\n\n // treat the bbox as a center point\n if (\n bboxList.length === 6 ||\n bboxList.length === 2 ||\n bboxList.length === 3 ||\n bboxList.length === 7\n ) {\n return [\n Math.max(\n 0,\n Math.round((bboxList[0] * width) / 1000) - defaultBboxSize / 2,\n ),\n Math.max(\n 0,\n Math.round((bboxList[1] * height) / 1000) - defaultBboxSize / 2,\n ),\n Math.min(\n width,\n Math.round((bboxList[0] * width) / 1000) + defaultBboxSize / 2,\n ),\n Math.min(\n height,\n Math.round((bboxList[1] * height) / 1000) + defaultBboxSize / 2,\n ),\n ];\n }\n\n if (bbox.length === 8) {\n return [\n Math.round((bboxList[0] * width) / 1000),\n Math.round((bboxList[1] * height) / 1000),\n Math.round((bboxList[4] * width) / 1000),\n Math.round((bboxList[5] * height) / 1000),\n ];\n }\n\n const msg = `invalid bbox data for doubao-vision mode: ${JSON.stringify(bbox)} `;\n throw new Error(msg);\n}\n\nexport function adaptBbox(\n bbox: number[],\n width: number,\n height: number,\n modelPreferences: IModelPreferences,\n): [number, number, number, number] {\n if (\n vlLocateMode(modelPreferences) === 'doubao-vision' ||\n vlLocateMode(modelPreferences) === 'vlm-ui-tars'\n ) {\n return adaptDoubaoBbox(bbox, width, height);\n }\n\n if (vlLocateMode(modelPreferences) === 'gemini') {\n return adaptGeminiBbox(bbox, width, height);\n }\n\n return adaptQwenBbox(bbox);\n}\n\nexport function adaptGeminiBbox(\n bbox: number[],\n width: number,\n height: number,\n): [number, number, number, number] {\n const left = Math.round((bbox[1] * width) / 1000);\n const top = Math.round((bbox[0] * height) / 1000);\n const right = Math.round((bbox[3] * width) / 1000);\n const bottom = Math.round((bbox[2] * height) / 1000);\n return [left, top, right, bottom];\n}\n\nexport function adaptBboxToRect(\n bbox: number[],\n width: number,\n height: number,\n modelPreferences: IModelPreferences,\n offsetX = 0,\n offsetY = 0,\n): Rect {\n debugInspectUtils('adaptBboxToRect', bbox, width, height, offsetX, offsetY);\n const [left, top, right, bottom] = adaptBbox(\n bbox,\n width,\n height,\n modelPreferences,\n );\n\n // Calculate initial rect dimensions\n const rectLeft = left;\n const rectTop = top;\n let rectWidth = right - left;\n let rectHeight = bottom - top;\n\n // Ensure the rect doesn't exceed image boundaries\n // If right edge exceeds width, adjust the width\n if (rectLeft + rectWidth > width) {\n rectWidth = width - rectLeft;\n }\n\n // If bottom edge exceeds height, adjust the height\n if (rectTop + rectHeight > height) {\n rectHeight = height - rectTop;\n }\n\n // Ensure minimum dimensions (width and height should be at least 1)\n rectWidth = Math.max(1, rectWidth);\n rectHeight = Math.max(1, rectHeight);\n\n const rect = {\n left: rectLeft + offsetX,\n top: rectTop + offsetY,\n width: rectWidth,\n height: rectHeight,\n };\n debugInspectUtils('adaptBboxToRect, result=', rect);\n return rect;\n}\n\nlet warned = false;\nexport function warnGPT4oSizeLimit(\n size: Size,\n modelPreferences: IModelPreferences,\n) {\n if (warned) return;\n if (getModelName(modelPreferences)?.toLowerCase().includes('gpt-4o')) {\n const warningMsg = `GPT-4o has a maximum image input size of 2000x768 or 768x2000, but got ${size.width}x${size.height}. Please set your interface to a smaller resolution. Otherwise, the result may be inaccurate.`;\n\n if (\n Math.max(size.width, size.height) > 2000 ||\n Math.min(size.width, size.height) > 768\n ) {\n console.warn(warningMsg);\n warned = true;\n }\n } else if (size.width > 1800 || size.height > 1800) {\n console.warn(\n `The image size seems too large (${size.width}x${size.height}). It may lead to more token usage, slower response, and inaccurate result.`,\n );\n warned = true;\n }\n}\n\nexport function mergeRects(rects: Rect[]) {\n const minLeft = Math.min(...rects.map((r) => r.left));\n const minTop = Math.min(...rects.map((r) => r.top));\n const maxRight = Math.max(...rects.map((r) => r.left + r.width));\n const maxBottom = Math.max(...rects.map((r) => r.top + r.height));\n return {\n left: minLeft,\n top: minTop,\n width: maxRight - minLeft,\n height: maxBottom - minTop,\n };\n}\n\n// expand the search area to at least 300 x 300, or add a default padding\nexport function expandSearchArea(\n rect: Rect,\n screenSize: Size,\n modelPreferences: IModelPreferences,\n) {\n const minEdgeSize =\n vlLocateMode(modelPreferences) === 'doubao-vision' ? 500 : 300;\n const defaultPadding = 160;\n\n const paddingSizeHorizontal =\n rect.width < minEdgeSize\n ? Math.ceil((minEdgeSize - rect.width) / 2)\n : defaultPadding;\n const paddingSizeVertical =\n rect.height < minEdgeSize\n ? Math.ceil((minEdgeSize - rect.height) / 2)\n : defaultPadding;\n rect.left = Math.max(0, rect.left - paddingSizeHorizontal);\n rect.width = Math.min(\n rect.width + paddingSizeHorizontal * 2,\n screenSize.width - rect.left,\n );\n rect.top = Math.max(0, rect.top - paddingSizeVertical);\n rect.height = Math.min(\n rect.height + paddingSizeVertical * 2,\n screenSize.height - rect.top,\n );\n return rect;\n}\n\nexport async function markupImageForLLM(\n screenshotBase64: string,\n tree: ElementTreeNode<BaseElement>,\n size: Size,\n) {\n const elementsInfo = treeToList(tree);\n const elementsPositionInfoWithoutText = elementsInfo!.filter(\n (elementInfo) => {\n if (elementInfo.attributes.nodeType === NodeType.TEXT) {\n return false;\n }\n return true;\n },\n );\n\n const imagePayload = await compositeElementInfoImg({\n inputImgBase64: screenshotBase64,\n elementsPositionInfo: elementsPositionInfoWithoutText,\n size,\n });\n return imagePayload;\n}\n\nexport function buildYamlFlowFromPlans(\n plans: PlanningAction[],\n actionSpace: DeviceAction<any>[],\n sleep?: number,\n): MidsceneYamlFlowItem[] {\n const flow: MidsceneYamlFlowItem[] = [];\n\n for (const plan of plans) {\n const verb = plan.type;\n\n const action = actionSpace.find((action) => action.name === verb);\n if (!action) {\n console.warn(\n `Cannot convert action ${verb} to yaml flow. Will ignore it.`,\n );\n continue;\n }\n\n const flowKey = action.interfaceAlias || verb;\n const flowParam = action.paramSchema\n ? dumpActionParam(plan.param || {}, action.paramSchema)\n : {};\n\n const flowItem: MidsceneYamlFlowItem = {\n [flowKey]: '',\n ...flowParam,\n };\n\n flow.push(flowItem);\n }\n\n if (sleep) {\n flow.push({\n sleep,\n });\n }\n\n return flow;\n}\n\n// Zod schemas for shared types\nexport const PointSchema = z.object({\n left: z.number(),\n top: z.number(),\n});\n\nexport const SizeSchema = z.object({\n width: z.number(),\n height: z.number(),\n dpr: z.number().optional(),\n});\n\nexport const RectSchema = PointSchema.and(SizeSchema).and(\n z.object({\n zoom: z.number().optional(),\n }),\n);\n\n// Zod schema for TMultimodalPrompt\nexport const TMultimodalPromptSchema = z.object({\n images: z\n .array(\n z.object({\n name: z.string(),\n url: z.string(),\n }),\n )\n .optional(),\n convertHttpImage2Base64: z.boolean().optional(),\n});\n\n// Zod schema for TUserPrompt\nexport const TUserPromptSchema = z.union([\n z.string(),\n z\n .object({\n prompt: z.string(),\n })\n .and(TMultimodalPromptSchema.partial()),\n]);\n\n// Generate TypeScript types from Zod schemas\nexport type TMultimodalPrompt = z.infer<typeof TMultimodalPromptSchema>;\nexport type TUserPrompt = z.infer<typeof TUserPromptSchema>;\n\nconst locateFieldFlagName = 'midscene_location_field_flag';\n\nconst MidsceneLocationResult = z\n .object({\n [locateFieldFlagName]: z.literal(true),\n prompt: TUserPromptSchema,\n\n // optional fields\n deepThink: z.boolean().optional(), // only available in vl model\n cacheable: z.boolean().optional(),\n xpath: z.boolean().optional(), // preset result for xpath\n\n // these two fields will only appear in the result\n center: z.tuple([z.number(), z.number()]),\n rect: RectSchema,\n })\n .passthrough();\n\nexport type MidsceneLocationResultType = z.infer<typeof MidsceneLocationResult>;\nexport const getMidsceneLocationSchema = () => {\n return MidsceneLocationResult;\n};\n\nexport const ifMidsceneLocatorField = (field: any): boolean => {\n // Handle optional fields by getting the inner type\n let actualField = field;\n if (actualField._def?.typeName === 'ZodOptional') {\n actualField = actualField._def.innerType;\n }\n\n // Check if this is a ZodUnion (the new MidsceneLocation structure)\n if (actualField._def?.typeName === 'ZodObject') {\n const shape = actualField._def.shape();\n return locateFieldFlagName in shape;\n }\n\n return false;\n};\n\nexport const dumpMidsceneLocatorField = (field: any): string => {\n assert(\n ifMidsceneLocatorField(field),\n 'field is not a midscene locator field',\n );\n\n // If field is a string, return it directly\n if (typeof field === 'string') {\n return field;\n }\n\n // If field is an object with prompt property\n if (field && typeof field === 'object' && field.prompt) {\n // If prompt is a string, return it directly\n if (typeof field.prompt === 'string') {\n return field.prompt;\n }\n // If prompt is a TUserPrompt object, extract the prompt string\n if (typeof field.prompt === 'object' && field.prompt.prompt) {\n return field.prompt.prompt; // TODO: dump images if necessary\n }\n }\n\n // Fallback: try to convert to string\n return String(field);\n};\n\nexport const findAllMidsceneLocatorField = (\n zodType?: z.ZodType<any>,\n requiredOnly?: boolean,\n): string[] => {\n if (!zodType) {\n return [];\n }\n\n // Check if this is a ZodObject by checking if it has a shape property\n const zodObject = zodType as any;\n if (zodObject._def?.typeName === 'ZodObject' && zodObject.shape) {\n const keys = Object.keys(zodObject.shape);\n return keys.filter((key) => {\n const field = zodObject.shape[key];\n if (!ifMidsceneLocatorField(field)) {\n return false;\n }\n\n // If requiredOnly is true, filter out optional fields\n if (requiredOnly) {\n return field._def?.typeName !== 'ZodOptional';\n }\n\n return true;\n });\n }\n\n // For other ZodType instances, we can't extract field names\n return [];\n};\n\nexport const dumpActionParam = (\n jsonObject: Record<string, any>,\n zodSchema: z.ZodType<any>,\n): Record<string, any> => {\n const locatorFields = findAllMidsceneLocatorField(zodSchema);\n const result = { ...jsonObject };\n\n for (const fieldName of locatorFields) {\n const fieldValue = result[fieldName];\n if (fieldValue) {\n // If it's already a string, keep it as is\n if (typeof fieldValue === 'string') {\n result[fieldName] = fieldValue;\n } else if (typeof fieldValue === 'object') {\n // Check if this field is actually a MidsceneLocationType object\n if (fieldValue.prompt) {\n // If prompt is a string, use it directly\n if (typeof fieldValue.prompt === 'string') {\n result[fieldName] = fieldValue.prompt;\n } else if (\n typeof fieldValue.prompt === 'object' &&\n fieldValue.prompt.prompt\n ) {\n // If prompt is a TUserPrompt object, extract the prompt string\n result[fieldName] = fieldValue.prompt.prompt;\n }\n }\n }\n }\n }\n\n return result;\n};\n\nexport const loadActionParam = (\n jsonObject: Record<string, any>,\n zodSchema: z.ZodType<any>,\n): Record<string, any> => {\n const locatorFields = findAllMidsceneLocatorField(zodSchema);\n const result = { ...jsonObject };\n\n for (const fieldName of locatorFields) {\n const fieldValue = result[fieldName];\n if (fieldValue && typeof fieldValue === 'string') {\n result[fieldName] = {\n [locateFieldFlagName]: true,\n prompt: fieldValue,\n };\n }\n }\n\n return result;\n};\n"],"names":["__webpack_require__","definition","key","Object","obj","prop","Symbol","AIActionType","callAiFn","msgs","AIActionTypeValue","modelPreferences","jsonObject","callToGetJSONObject","defaultBboxSize","debugInspectUtils","getDebug","fillBboxParam","locate","width","height","adaptBbox","adaptQwenBbox","bbox","msg","JSON","Error","result","Math","adaptDoubaoBbox","assert","splitted","Number","Array","bboxList","item","x","y","vlLocateMode","adaptGeminiBbox","left","top","right","bottom","adaptBboxToRect","offsetX","offsetY","rectLeft","rectTop","rectWidth","rectHeight","rect","warned","warnGPT4oSizeLimit","size","_getModelName","warningMsg","console","mergeRects","rects","minLeft","r","minTop","maxRight","maxBottom","expandSearchArea","screenSize","minEdgeSize","defaultPadding","paddingSizeHorizontal","paddingSizeVertical","markupImageForLLM","screenshotBase64","tree","elementsInfo","treeToList","elementsPositionInfoWithoutText","elementInfo","NodeType","imagePayload","compositeElementInfoImg","buildYamlFlowFromPlans","plans","actionSpace","sleep","flow","plan","verb","action","flowKey","flowParam","dumpActionParam","flowItem","PointSchema","z","SizeSchema","RectSchema","TMultimodalPromptSchema","TUserPromptSchema","locateFieldFlagName","MidsceneLocationResult","getMidsceneLocationSchema","ifMidsceneLocatorField","field","_actualField__def","_actualField__def1","actualField","shape","dumpMidsceneLocatorField","String","findAllMidsceneLocatorField","zodType","requiredOnly","_zodObject__def","zodObject","keys","_field__def","zodSchema","locatorFields","fieldName","fieldValue","loadActionParam"],"mappings":";;;IAAAA,oBAAoB,CAAC,GAAG,CAAC,UAASC;QACjC,IAAI,IAAIC,OAAOD,WACR,IAAGD,oBAAoB,CAAC,CAACC,YAAYC,QAAQ,CAACF,oBAAoB,CAAC,CAAC,UAASE,MACzEC,OAAO,cAAc,CAAC,UAASD,KAAK;YAAE,YAAY;YAAM,KAAKD,UAAU,CAACC,IAAI;QAAC;IAGzF;;;ICNAF,oBAAoB,CAAC,GAAG,CAACI,KAAKC,OAAUF,OAAO,SAAS,CAAC,cAAc,CAAC,IAAI,CAACC,KAAKC;;;ICClFL,oBAAoB,CAAC,GAAG,CAAC;QACxB,IAAG,AAAkB,eAAlB,OAAOM,UAA0BA,OAAO,WAAW,EACrDH,OAAO,cAAc,CAAC,UAASG,OAAO,WAAW,EAAE;YAAE,OAAO;QAAS;QAEtEH,OAAO,cAAc,CAAC,UAAS,cAAc;YAAE,OAAO;QAAK;IAC5D;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AC6BO,IAAKI,sBAAYA,WAAAA,GAAAA,SAAZA,YAAY;;;;;;WAAZA;;AAQL,eAAeC,SACpBC,IAAY,EACZC,iBAA+B,EAC/BC,gBAAmC;IAEnC,MAAMC,aAAa,MAAMC,AAAAA,IAAAA,yBAAAA,mBAAAA,AAAAA,EACvBJ,MACAC,mBACAC;IAGF,OAAO;QACL,SAASC,WAAW,OAAO;QAC3B,OAAOA,WAAW,KAAK;IACzB;AACF;AAEA,MAAME,kBAAkB;AACxB,MAAMC,oBAAoBC,AAAAA,IAAAA,uBAAAA,QAAAA,AAAAA,EAAS;AAG5B,SAASC,cACdC,MAA2B,EAC3BC,KAAa,EACbC,MAAc,EACdT,gBAAmC;IAGnC,IAAKO,OAAe,OAAO,IAAI,CAACA,CAAAA,QAAAA,SAAAA,KAAAA,IAAAA,OAAQ,IAAI,AAAD,GAAG;QAC5CA,OAAO,IAAI,GAAIA,OAAe,OAAO;QAErC,OAAQA,OAAe,OAAO;IAChC;IAEA,IAAIA,QAAAA,SAAAA,KAAAA,IAAAA,OAAQ,IAAI,EACdA,OAAO,IAAI,GAAGG,UAAUH,OAAO,IAAI,EAAEC,OAAOC,QAAQT;IAGtD,OAAOO;AACT;AAEO,SAASI,cACdC,IAAc;IAEd,IAAIA,KAAK,MAAM,GAAG,GAAG;QACnB,MAAMC,MAAM,CAAC,oCAAoC,EAAEC,KAAK,SAAS,CAACF,MAAM,CAAC,CAAC;QAC1E,MAAM,IAAIG,MAAMF;IAClB;IAEA,MAAMG,SAA2C;QAC/CC,KAAK,KAAK,CAACL,IAAI,CAAC,EAAE;QAClBK,KAAK,KAAK,CAACL,IAAI,CAAC,EAAE;QACC,YAAnB,OAAOA,IAAI,CAAC,EAAE,GACVK,KAAK,KAAK,CAACL,IAAI,CAAC,EAAE,IAClBK,KAAK,KAAK,CAACL,IAAI,CAAC,EAAE,GAAGT;QACN,YAAnB,OAAOS,IAAI,CAAC,EAAE,GACVK,KAAK,KAAK,CAACL,IAAI,CAAC,EAAE,IAClBK,KAAK,KAAK,CAACL,IAAI,CAAC,EAAE,GAAGT;KAC1B;IACD,OAAOa;AACT;AAEO,SAASE,gBACdN,IAAkC,EAClCJ,KAAa,EACbC,MAAc;IAEdU,IAAAA,sBAAAA,MAAAA,AAAAA,EACEX,QAAQ,KAAKC,SAAS,GACtB;IAGF,IAAI,AAAgB,YAAhB,OAAOG,MAAmB;QAC5BO,IAAAA,sBAAAA,MAAAA,AAAAA,EACE,+BAA+B,IAAI,CAACP,KAAK,IAAI,KAC7C,CAAC,iDAAiD,EAAEA,MAAM;QAE5D,MAAMQ,WAAWR,KAAK,KAAK,CAAC;QAC5B,IAAIQ,AAAoB,MAApBA,SAAS,MAAM,EACjB,OAAO;YACLH,KAAK,KAAK,CAAEI,OAAOD,QAAQ,CAAC,EAAE,IAAIZ,QAAS;YAC3CS,KAAK,KAAK,CAAEI,OAAOD,QAAQ,CAAC,EAAE,IAAIX,SAAU;YAC5CQ,KAAK,KAAK,CAAEI,OAAOD,QAAQ,CAAC,EAAE,IAAIZ,QAAS;YAC3CS,KAAK,KAAK,CAAEI,OAAOD,QAAQ,CAAC,EAAE,IAAIX,SAAU;SAC7C;QAEH,MAAM,IAAIM,MAAM,CAAC,iDAAiD,EAAEH,MAAM;IAC5E;IAEA,IAAIU,MAAM,OAAO,CAACV,SAASU,MAAM,OAAO,CAACV,IAAI,CAAC,EAAE,GAC9CA,OAAOA,IAAI,CAAC,EAAE;IAGhB,IAAIW,WAAqB,EAAE;IAC3B,IAAID,MAAM,OAAO,CAACV,SAAS,AAAmB,YAAnB,OAAOA,IAAI,CAAC,EAAE,EACvCA,KAAK,OAAO,CAAC,CAACY;QACZ,IAAI,AAAgB,YAAhB,OAAOA,QAAqBA,KAAK,QAAQ,CAAC,MAAM;YAClD,MAAM,CAACC,GAAGC,EAAE,GAAGF,KAAK,KAAK,CAAC;YAC1BD,SAAS,IAAI,CAACF,OAAOI,EAAE,IAAI,KAAKJ,OAAOK,EAAE,IAAI;QAC/C,OAAO,IAAI,AAAgB,YAAhB,OAAOF,QAAqBA,KAAK,QAAQ,CAAC,MAAM;YACzD,MAAM,CAACC,GAAGC,EAAE,GAAGF,KAAK,KAAK,CAAC;YAC1BD,SAAS,IAAI,CAACF,OAAOI,EAAE,IAAI,KAAKJ,OAAOK,EAAE,IAAI;QAC/C,OACEH,SAAS,IAAI,CAACF,OAAOG;IAEzB;SAEAD,WAAWX;IAGb,IAAIW,AAAoB,MAApBA,SAAS,MAAM,IAAUA,AAAoB,MAApBA,SAAS,MAAM,EAC1C,OAAO;QACLN,KAAK,KAAK,CAAEM,QAAQ,CAAC,EAAE,GAAGf,QAAS;QACnCS,KAAK,KAAK,CAAEM,QAAQ,CAAC,EAAE,GAAGd,SAAU;QACpCQ,KAAK,KAAK,CAAEM,QAAQ,CAAC,EAAE,GAAGf,QAAS;QACnCS,KAAK,KAAK,CAAEM,QAAQ,CAAC,EAAE,GAAGd,SAAU;KACrC;IAIH,IACEc,AAAoB,MAApBA,SAAS,MAAM,IACfA,AAAoB,MAApBA,SAAS,MAAM,IACfA,AAAoB,MAApBA,SAAS,MAAM,IACfA,AAAoB,MAApBA,SAAS,MAAM,EAEf,OAAO;QACLN,KAAK,GAAG,CACN,GACAA,KAAK,KAAK,CAAEM,QAAQ,CAAC,EAAE,GAAGf,QAAS,QAAQL,kBAAkB;QAE/Dc,KAAK,GAAG,CACN,GACAA,KAAK,KAAK,CAAEM,QAAQ,CAAC,EAAE,GAAGd,SAAU,QAAQN,kBAAkB;QAEhEc,KAAK,GAAG,CACNT,OACAS,KAAK,KAAK,CAAEM,QAAQ,CAAC,EAAE,GAAGf,QAAS,QAAQL,kBAAkB;QAE/Dc,KAAK,GAAG,CACNR,QACAQ,KAAK,KAAK,CAAEM,QAAQ,CAAC,EAAE,GAAGd,SAAU,QAAQN,kBAAkB;KAEjE;IAGH,IAAIS,AAAgB,MAAhBA,KAAK,MAAM,EACb,OAAO;QACLK,KAAK,KAAK,CAAEM,QAAQ,CAAC,EAAE,GAAGf,QAAS;QACnCS,KAAK,KAAK,CAAEM,QAAQ,CAAC,EAAE,GAAGd,SAAU;QACpCQ,KAAK,KAAK,CAAEM,QAAQ,CAAC,EAAE,GAAGf,QAAS;QACnCS,KAAK,KAAK,CAAEM,QAAQ,CAAC,EAAE,GAAGd,SAAU;KACrC;IAGH,MAAMI,MAAM,CAAC,0CAA0C,EAAEC,KAAK,SAAS,CAACF,MAAM,CAAC,CAAC;IAChF,MAAM,IAAIG,MAAMF;AAClB;AAEO,SAASH,UACdE,IAAc,EACdJ,KAAa,EACbC,MAAc,EACdT,gBAAmC;IAEnC,IACE2B,AAAmC,oBAAnCA,AAAAA,IAAAA,oBAAAA,YAAAA,AAAAA,EAAa3B,qBACb2B,AAAmC,kBAAnCA,AAAAA,IAAAA,oBAAAA,YAAAA,AAAAA,EAAa3B,mBAEb,OAAOkB,gBAAgBN,MAAMJ,OAAOC;IAGtC,IAAIkB,AAAmC,aAAnCA,AAAAA,IAAAA,oBAAAA,YAAAA,AAAAA,EAAa3B,mBACf,OAAO4B,gBAAgBhB,MAAMJ,OAAOC;IAGtC,OAAOE,cAAcC;AACvB;AAEO,SAASgB,gBACdhB,IAAc,EACdJ,KAAa,EACbC,MAAc;IAEd,MAAMoB,OAAOZ,KAAK,KAAK,CAAEL,IAAI,CAAC,EAAE,GAAGJ,QAAS;IAC5C,MAAMsB,MAAMb,KAAK,KAAK,CAAEL,IAAI,CAAC,EAAE,GAAGH,SAAU;IAC5C,MAAMsB,QAAQd,KAAK,KAAK,CAAEL,IAAI,CAAC,EAAE,GAAGJ,QAAS;IAC7C,MAAMwB,SAASf,KAAK,KAAK,CAAEL,IAAI,CAAC,EAAE,GAAGH,SAAU;IAC/C,OAAO;QAACoB;QAAMC;QAAKC;QAAOC;KAAO;AACnC;AAEO,SAASC,gBACdrB,IAAc,EACdJ,KAAa,EACbC,MAAc,EACdT,gBAAmC,EACnCkC,UAAU,CAAC,EACXC,UAAU,CAAC;IAEX/B,kBAAkB,mBAAmBQ,MAAMJ,OAAOC,QAAQyB,SAASC;IACnE,MAAM,CAACN,MAAMC,KAAKC,OAAOC,OAAO,GAAGtB,UACjCE,MACAJ,OACAC,QACAT;IAIF,MAAMoC,WAAWP;IACjB,MAAMQ,UAAUP;IAChB,IAAIQ,YAAYP,QAAQF;IACxB,IAAIU,aAAaP,SAASF;IAI1B,IAAIM,WAAWE,YAAY9B,OACzB8B,YAAY9B,QAAQ4B;IAItB,IAAIC,UAAUE,aAAa9B,QACzB8B,aAAa9B,SAAS4B;IAIxBC,YAAYrB,KAAK,GAAG,CAAC,GAAGqB;IACxBC,aAAatB,KAAK,GAAG,CAAC,GAAGsB;IAEzB,MAAMC,OAAO;QACX,MAAMJ,WAAWF;QACjB,KAAKG,UAAUF;QACf,OAAOG;QACP,QAAQC;IACV;IACAnC,kBAAkB,4BAA4BoC;IAC9C,OAAOA;AACT;AAEA,IAAIC,SAAS;AACN,SAASC,mBACdC,IAAU,EACV3C,gBAAmC;QAG/B4C;IADJ,IAAIH,QAAQ;IACZ,IAAI,QAAAG,CAAAA,gBAAAA,AAAAA,IAAAA,oBAAAA,YAAAA,AAAAA,EAAa5C,iBAAgB,IAA7B4C,KAAAA,IAAAA,cAAgC,WAAW,GAAG,QAAQ,CAAC,WAAW;QACpE,MAAMC,aAAa,CAAC,uEAAuE,EAAEF,KAAK,KAAK,CAAC,CAAC,EAAEA,KAAK,MAAM,CAAC,6FAA6F,CAAC;QAErN,IACE1B,KAAK,GAAG,CAAC0B,KAAK,KAAK,EAAEA,KAAK,MAAM,IAAI,QACpC1B,KAAK,GAAG,CAAC0B,KAAK,KAAK,EAAEA,KAAK,MAAM,IAAI,KACpC;YACAG,QAAQ,IAAI,CAACD;YACbJ,SAAS;QACX;IACF,OAAO,IAAIE,KAAK,KAAK,GAAG,QAAQA,KAAK,MAAM,GAAG,MAAM;QAClDG,QAAQ,IAAI,CACV,CAAC,gCAAgC,EAAEH,KAAK,KAAK,CAAC,CAAC,EAAEA,KAAK,MAAM,CAAC,2EAA2E,CAAC;QAE3IF,SAAS;IACX;AACF;AAEO,SAASM,WAAWC,KAAa;IACtC,MAAMC,UAAUhC,KAAK,GAAG,IAAI+B,MAAM,GAAG,CAAC,CAACE,IAAMA,EAAE,IAAI;IACnD,MAAMC,SAASlC,KAAK,GAAG,IAAI+B,MAAM,GAAG,CAAC,CAACE,IAAMA,EAAE,GAAG;IACjD,MAAME,WAAWnC,KAAK,GAAG,IAAI+B,MAAM,GAAG,CAAC,CAACE,IAAMA,EAAE,IAAI,GAAGA,EAAE,KAAK;IAC9D,MAAMG,YAAYpC,KAAK,GAAG,IAAI+B,MAAM,GAAG,CAAC,CAACE,IAAMA,EAAE,GAAG,GAAGA,EAAE,MAAM;IAC/D,OAAO;QACL,MAAMD;QACN,KAAKE;QACL,OAAOC,WAAWH;QAClB,QAAQI,YAAYF;IACtB;AACF;AAGO,SAASG,iBACdd,IAAU,EACVe,UAAgB,EAChBvD,gBAAmC;IAEnC,MAAMwD,cACJ7B,AAAmC,oBAAnCA,AAAAA,IAAAA,oBAAAA,YAAAA,AAAAA,EAAa3B,oBAAwC,MAAM;IAC7D,MAAMyD,iBAAiB;IAEvB,MAAMC,wBACJlB,KAAK,KAAK,GAAGgB,cACTvC,KAAK,IAAI,CAAEuC,AAAAA,CAAAA,cAAchB,KAAK,KAAI,IAAK,KACvCiB;IACN,MAAME,sBACJnB,KAAK,MAAM,GAAGgB,cACVvC,KAAK,IAAI,CAAEuC,AAAAA,CAAAA,cAAchB,KAAK,MAAK,IAAK,KACxCiB;IACNjB,KAAK,IAAI,GAAGvB,KAAK,GAAG,CAAC,GAAGuB,KAAK,IAAI,GAAGkB;IACpClB,KAAK,KAAK,GAAGvB,KAAK,GAAG,CACnBuB,KAAK,KAAK,GAAGkB,AAAwB,IAAxBA,uBACbH,WAAW,KAAK,GAAGf,KAAK,IAAI;IAE9BA,KAAK,GAAG,GAAGvB,KAAK,GAAG,CAAC,GAAGuB,KAAK,GAAG,GAAGmB;IAClCnB,KAAK,MAAM,GAAGvB,KAAK,GAAG,CACpBuB,KAAK,MAAM,GAAGmB,AAAsB,IAAtBA,qBACdJ,WAAW,MAAM,GAAGf,KAAK,GAAG;IAE9B,OAAOA;AACT;AAEO,eAAeoB,kBACpBC,gBAAwB,EACxBC,IAAkC,EAClCnB,IAAU;IAEV,MAAMoB,eAAeC,AAAAA,IAAAA,0BAAAA,UAAAA,AAAAA,EAAWF;IAChC,MAAMG,kCAAkCF,aAAc,MAAM,CAC1D,CAACG;QACC,IAAIA,YAAY,UAAU,CAAC,QAAQ,KAAKC,0BAAAA,QAAAA,CAAAA,IAAa,EACnD,OAAO;QAET,OAAO;IACT;IAGF,MAAMC,eAAe,MAAMC,AAAAA,IAAAA,oBAAAA,uBAAAA,AAAAA,EAAwB;QACjD,gBAAgBR;QAChB,sBAAsBI;QACtBtB;IACF;IACA,OAAOyB;AACT;AAEO,SAASE,uBACdC,KAAuB,EACvBC,WAAgC,EAChCC,KAAc;IAEd,MAAMC,OAA+B,EAAE;IAEvC,KAAK,MAAMC,QAAQJ,MAAO;QACxB,MAAMK,OAAOD,KAAK,IAAI;QAEtB,MAAME,SAASL,YAAY,IAAI,CAAC,CAACK,SAAWA,OAAO,IAAI,KAAKD;QAC5D,IAAI,CAACC,QAAQ;YACX/B,QAAQ,IAAI,CACV,CAAC,sBAAsB,EAAE8B,KAAK,8BAA8B,CAAC;YAE/D;QACF;QAEA,MAAME,UAAUD,OAAO,cAAc,IAAID;QACzC,MAAMG,YAAYF,OAAO,WAAW,GAChCG,gBAAgBL,KAAK,KAAK,IAAI,CAAC,GAAGE,OAAO,WAAW,IACpD,CAAC;QAEL,MAAMI,WAAiC;YACrC,CAACH,QAAQ,EAAE;YACX,GAAGC,SAAS;QACd;QAEAL,KAAK,IAAI,CAACO;IACZ;IAEA,IAAIR,OACFC,KAAK,IAAI,CAAC;QACRD;IACF;IAGF,OAAOC;AACT;AAGO,MAAMQ,cAAcC,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IAClC,MAAMA,6BAAAA,CAAAA,CAAAA,MAAQ;IACd,KAAKA,6BAAAA,CAAAA,CAAAA,MAAQ;AACf;AAEO,MAAMC,aAAaD,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IACjC,OAAOA,6BAAAA,CAAAA,CAAAA,MAAQ;IACf,QAAQA,6BAAAA,CAAAA,CAAAA,MAAQ;IAChB,KAAKA,6BAAAA,CAAAA,CAAAA,MAAQ,GAAG,QAAQ;AAC1B;AAEO,MAAME,aAAaH,YAAY,GAAG,CAACE,YAAY,GAAG,CACvDD,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IACP,MAAMA,6BAAAA,CAAAA,CAAAA,MAAQ,GAAG,QAAQ;AAC3B;AAIK,MAAMG,0BAA0BH,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IAC9C,QAAQA,6BAAAA,CAAAA,CAAAA,KACA,CACJA,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;QACP,MAAMA,6BAAAA,CAAAA,CAAAA,MAAQ;QACd,KAAKA,6BAAAA,CAAAA,CAAAA,MAAQ;IACf,IAED,QAAQ;IACX,yBAAyBA,6BAAAA,CAAAA,CAAAA,OAAS,GAAG,QAAQ;AAC/C;AAGO,MAAMI,oBAAoBJ,6BAAAA,CAAAA,CAAAA,KAAO,CAAC;IACvCA,6BAAAA,CAAAA,CAAAA,MAAQ;IACRA,6BAAAA,CAAAA,CAAAA,MACS,CAAC;QACN,QAAQA,6BAAAA,CAAAA,CAAAA,MAAQ;IAClB,GACC,GAAG,CAACG,wBAAwB,OAAO;CACvC;AAMD,MAAME,sBAAsB;AAE5B,MAAMC,yBAAyBN,6BAAAA,CAAAA,CAAAA,MACtB,CAAC;IACN,CAACK,oBAAoB,EAAEL,6BAAAA,CAAAA,CAAAA,OAAS,CAAC;IACjC,QAAQI;IAGR,WAAWJ,6BAAAA,CAAAA,CAAAA,OAAS,GAAG,QAAQ;IAC/B,WAAWA,6BAAAA,CAAAA,CAAAA,OAAS,GAAG,QAAQ;IAC/B,OAAOA,6BAAAA,CAAAA,CAAAA,OAAS,GAAG,QAAQ;IAG3B,QAAQA,6BAAAA,CAAAA,CAAAA,KAAO,CAAC;QAACA,6BAAAA,CAAAA,CAAAA,MAAQ;QAAIA,6BAAAA,CAAAA,CAAAA,MAAQ;KAAG;IACxC,MAAME;AACR,GACC,WAAW;AAGP,MAAMK,4BAA4B,IAChCD;AAGF,MAAME,yBAAyB,CAACC;QAGjCC,mBAKAC;IANJ,IAAIC,cAAcH;IAClB,IAAIC,AAAAA,SAAAA,CAAAA,oBAAAA,YAAY,IAAI,AAAD,IAAfA,KAAAA,IAAAA,kBAAkB,QAAQ,AAAD,MAAM,eACjCE,cAAcA,YAAY,IAAI,CAAC,SAAS;IAI1C,IAAID,AAAAA,SAAAA,CAAAA,qBAAAA,YAAY,IAAI,AAAD,IAAfA,KAAAA,IAAAA,mBAAkB,QAAQ,AAAD,MAAM,aAAa;QAC9C,MAAME,QAAQD,YAAY,IAAI,CAAC,KAAK;QACpC,OAAOP,uBAAuBQ;IAChC;IAEA,OAAO;AACT;AAEO,MAAMC,2BAA2B,CAACL;IACvCzE,IAAAA,sBAAAA,MAAAA,AAAAA,EACEwE,uBAAuBC,QACvB;IAIF,IAAI,AAAiB,YAAjB,OAAOA,OACT,OAAOA;IAIT,IAAIA,SAAS,AAAiB,YAAjB,OAAOA,SAAsBA,MAAM,MAAM,EAAE;QAEtD,IAAI,AAAwB,YAAxB,OAAOA,MAAM,MAAM,EACrB,OAAOA,MAAM,MAAM;QAGrB,IAAI,AAAwB,YAAxB,OAAOA,MAAM,MAAM,IAAiBA,MAAM,MAAM,CAAC,MAAM,EACzD,OAAOA,MAAM,MAAM,CAAC,MAAM;IAE9B;IAGA,OAAOM,OAAON;AAChB;AAEO,MAAMO,8BAA8B,CACzCC,SACAC;QAQIC;IANJ,IAAI,CAACF,SACH,OAAO,EAAE;IAIX,MAAMG,YAAYH;IAClB,IAAIE,AAAAA,SAAAA,CAAAA,kBAAAA,UAAU,IAAI,AAAD,IAAbA,KAAAA,IAAAA,gBAAgB,QAAQ,AAAD,MAAM,eAAeC,UAAU,KAAK,EAAE;QAC/D,MAAMC,OAAOhH,OAAO,IAAI,CAAC+G,UAAU,KAAK;QACxC,OAAOC,KAAK,MAAM,CAAC,CAACjH;YAClB,MAAMqG,QAAQW,UAAU,KAAK,CAAChH,IAAI;YAClC,IAAI,CAACoG,uBAAuBC,QAC1B,OAAO;YAIT,IAAIS,cAAc;oBACTI;gBAAP,OAAOA,AAAAA,SAAAA,CAAAA,cAAAA,MAAM,IAAI,AAAD,IAATA,KAAAA,IAAAA,YAAY,QAAQ,AAAD,MAAM;YAClC;YAEA,OAAO;QACT;IACF;IAGA,OAAO,EAAE;AACX;AAEO,MAAMzB,kBAAkB,CAC7B/E,YACAyG;IAEA,MAAMC,gBAAgBR,4BAA4BO;IAClD,MAAM1F,SAAS;QAAE,GAAGf,UAAU;IAAC;IAE/B,KAAK,MAAM2G,aAAaD,cAAe;QACrC,MAAME,aAAa7F,MAAM,CAAC4F,UAAU;QACpC,IAAIC,YAEF;YAAA,IAAI,AAAsB,YAAtB,OAAOA,YACT7F,MAAM,CAAC4F,UAAU,GAAGC;iBACf,IAAI,AAAsB,YAAtB,OAAOA,YAEhB;gBAAA,IAAIA,WAAW,MAAM,EAEnB;oBAAA,IAAI,AAA6B,YAA7B,OAAOA,WAAW,MAAM,EAC1B7F,MAAM,CAAC4F,UAAU,GAAGC,WAAW,MAAM;yBAChC,IACL,AAA6B,YAA7B,OAAOA,WAAW,MAAM,IACxBA,WAAW,MAAM,CAAC,MAAM,EAGxB7F,MAAM,CAAC4F,UAAU,GAAGC,WAAW,MAAM,CAAC,MAAM;gBAC9C;YACF;QACF;IAEJ;IAEA,OAAO7F;AACT;AAEO,MAAM8F,kBAAkB,CAC7B7G,YACAyG;IAEA,MAAMC,gBAAgBR,4BAA4BO;IAClD,MAAM1F,SAAS;QAAE,GAAGf,UAAU;IAAC;IAE/B,KAAK,MAAM2G,aAAaD,cAAe;QACrC,MAAME,aAAa7F,MAAM,CAAC4F,UAAU;QACpC,IAAIC,cAAc,AAAsB,YAAtB,OAAOA,YACvB7F,MAAM,CAAC4F,UAAU,GAAG;YAClB,CAACpB,oBAAoB,EAAE;YACvB,QAAQqB;QACV;IAEJ;IAEA,OAAO7F;AACT"}
1
+ {"version":3,"file":"ai-model/common.js","sources":["webpack://@midscene/core/webpack/runtime/define_property_getters","webpack://@midscene/core/webpack/runtime/has_own_property","webpack://@midscene/core/webpack/runtime/make_namespace_object","webpack://@midscene/core/./src/ai-model/common.ts"],"sourcesContent":["__webpack_require__.d = (exports, definition) => {\n\tfor(var key in definition) {\n if(__webpack_require__.o(definition, key) && !__webpack_require__.o(exports, key)) {\n Object.defineProperty(exports, key, { enumerable: true, get: definition[key] });\n }\n }\n};","__webpack_require__.o = (obj, prop) => (Object.prototype.hasOwnProperty.call(obj, prop))","// define __esModule on exports\n__webpack_require__.r = (exports) => {\n\tif(typeof Symbol !== 'undefined' && Symbol.toStringTag) {\n\t\tObject.defineProperty(exports, Symbol.toStringTag, { value: 'Module' });\n\t}\n\tObject.defineProperty(exports, '__esModule', { value: true });\n};","import type {\n AIUsageInfo,\n BaseElement,\n DeviceAction,\n ElementTreeNode,\n MidsceneYamlFlowItem,\n PlanningAction,\n Rect,\n Size,\n} from '@/types';\nimport { assert } from '@midscene/shared/utils';\n\nimport type { ChatCompletionMessageParam } from 'openai/resources/index';\n\nimport type { PlanningLocateParam } from '@/types';\nimport { NodeType } from '@midscene/shared/constants';\nimport type { TVlModeTypes } from '@midscene/shared/env';\nimport { treeToList } from '@midscene/shared/extractor';\nimport { compositeElementInfoImg } from '@midscene/shared/img';\nimport { getDebug } from '@midscene/shared/logger';\nimport { z } from 'zod';\n\nexport type AIArgs = ChatCompletionMessageParam[];\n\nexport enum AIActionType {\n ASSERT = 0,\n INSPECT_ELEMENT = 1,\n EXTRACT_DATA = 2,\n PLAN = 3,\n DESCRIBE_ELEMENT = 4,\n}\n\nconst defaultBboxSize = 20; // must be even number\nconst debugInspectUtils = getDebug('ai:common');\n\n// transform the param of locate from qwen mode\nexport function fillBboxParam(\n locate: PlanningLocateParam,\n width: number,\n height: number,\n vlMode: TVlModeTypes | undefined,\n) {\n // The Qwen model might have hallucinations of naming bbox as bbox_2d.\n if ((locate as any).bbox_2d && !locate?.bbox) {\n locate.bbox = (locate as any).bbox_2d;\n // biome-ignore lint/performance/noDelete: <explanation>\n delete (locate as any).bbox_2d;\n }\n\n if (locate?.bbox) {\n locate.bbox = adaptBbox(locate.bbox, width, height, vlMode);\n }\n\n return locate;\n}\n\nexport function adaptQwenBbox(\n bbox: number[],\n): [number, number, number, number] {\n if (bbox.length < 2) {\n const msg = `invalid bbox data for qwen-vl mode: ${JSON.stringify(bbox)} `;\n throw new Error(msg);\n }\n\n const result: [number, number, number, number] = [\n Math.round(bbox[0]),\n Math.round(bbox[1]),\n typeof bbox[2] === 'number'\n ? Math.round(bbox[2])\n : Math.round(bbox[0] + defaultBboxSize),\n typeof bbox[3] === 'number'\n ? Math.round(bbox[3])\n : Math.round(bbox[1] + defaultBboxSize),\n ];\n return result;\n}\n\nexport function adaptDoubaoBbox(\n bbox: string[] | number[] | string,\n width: number,\n height: number,\n): [number, number, number, number] {\n assert(\n width > 0 && height > 0,\n 'width and height must be greater than 0 in doubao mode',\n );\n\n if (typeof bbox === 'string') {\n assert(\n /^(\\d+)\\s(\\d+)\\s(\\d+)\\s(\\d+)$/.test(bbox.trim()),\n `invalid bbox data string for doubao-vision mode: ${bbox}`,\n );\n const splitted = bbox.split(' ');\n if (splitted.length === 4) {\n return [\n Math.round((Number(splitted[0]) * width) / 1000),\n Math.round((Number(splitted[1]) * height) / 1000),\n Math.round((Number(splitted[2]) * width) / 1000),\n Math.round((Number(splitted[3]) * height) / 1000),\n ];\n }\n throw new Error(`invalid bbox data string for doubao-vision mode: ${bbox}`);\n }\n\n if (Array.isArray(bbox) && Array.isArray(bbox[0])) {\n bbox = bbox[0];\n }\n\n let bboxList: number[] = [];\n if (Array.isArray(bbox) && typeof bbox[0] === 'string') {\n bbox.forEach((item) => {\n if (typeof item === 'string' && item.includes(',')) {\n const [x, y] = item.split(',');\n bboxList.push(Number(x.trim()), Number(y.trim()));\n } else if (typeof item === 'string' && item.includes(' ')) {\n const [x, y] = item.split(' ');\n bboxList.push(Number(x.trim()), Number(y.trim()));\n } else {\n bboxList.push(Number(item));\n }\n });\n } else {\n bboxList = bbox as any;\n }\n\n if (bboxList.length === 4 || bboxList.length === 5) {\n return [\n Math.round((bboxList[0] * width) / 1000),\n Math.round((bboxList[1] * height) / 1000),\n Math.round((bboxList[2] * width) / 1000),\n Math.round((bboxList[3] * height) / 1000),\n ];\n }\n\n // treat the bbox as a center point\n if (\n bboxList.length === 6 ||\n bboxList.length === 2 ||\n bboxList.length === 3 ||\n bboxList.length === 7\n ) {\n return [\n Math.max(\n 0,\n Math.round((bboxList[0] * width) / 1000) - defaultBboxSize / 2,\n ),\n Math.max(\n 0,\n Math.round((bboxList[1] * height) / 1000) - defaultBboxSize / 2,\n ),\n Math.min(\n width,\n Math.round((bboxList[0] * width) / 1000) + defaultBboxSize / 2,\n ),\n Math.min(\n height,\n Math.round((bboxList[1] * height) / 1000) + defaultBboxSize / 2,\n ),\n ];\n }\n\n if (bbox.length === 8) {\n return [\n Math.round((bboxList[0] * width) / 1000),\n Math.round((bboxList[1] * height) / 1000),\n Math.round((bboxList[4] * width) / 1000),\n Math.round((bboxList[5] * height) / 1000),\n ];\n }\n\n const msg = `invalid bbox data for doubao-vision mode: ${JSON.stringify(bbox)} `;\n throw new Error(msg);\n}\n\nexport function adaptBbox(\n bbox: number[],\n width: number,\n height: number,\n vlMode: TVlModeTypes | undefined,\n): [number, number, number, number] {\n if (vlMode === 'doubao-vision' || vlMode === 'vlm-ui-tars') {\n return adaptDoubaoBbox(bbox, width, height);\n }\n\n if (vlMode === 'gemini') {\n return adaptGeminiBbox(bbox, width, height);\n }\n\n return adaptQwenBbox(bbox);\n}\n\nexport function adaptGeminiBbox(\n bbox: number[],\n width: number,\n height: number,\n): [number, number, number, number] {\n const left = Math.round((bbox[1] * width) / 1000);\n const top = Math.round((bbox[0] * height) / 1000);\n const right = Math.round((bbox[3] * width) / 1000);\n const bottom = Math.round((bbox[2] * height) / 1000);\n return [left, top, right, bottom];\n}\n\nexport function adaptBboxToRect(\n bbox: number[],\n width: number,\n height: number,\n offsetX = 0,\n offsetY = 0,\n vlMode?: TVlModeTypes | undefined,\n): Rect {\n debugInspectUtils('adaptBboxToRect', bbox, width, height, offsetX, offsetY);\n const [left, top, right, bottom] = adaptBbox(bbox, width, height, vlMode);\n\n // Calculate initial rect dimensions\n const rectLeft = left;\n const rectTop = top;\n let rectWidth = right - left;\n let rectHeight = bottom - top;\n\n // Ensure the rect doesn't exceed image boundaries\n // If right edge exceeds width, adjust the width\n if (rectLeft + rectWidth > width) {\n rectWidth = width - rectLeft;\n }\n\n // If bottom edge exceeds height, adjust the height\n if (rectTop + rectHeight > height) {\n rectHeight = height - rectTop;\n }\n\n // Ensure minimum dimensions (width and height should be at least 1)\n rectWidth = Math.max(1, rectWidth);\n rectHeight = Math.max(1, rectHeight);\n\n const rect = {\n left: rectLeft + offsetX,\n top: rectTop + offsetY,\n width: rectWidth,\n height: rectHeight,\n };\n debugInspectUtils('adaptBboxToRect, result=', rect);\n return rect;\n}\n\nlet warned = false;\nexport function warnGPT4oSizeLimit(size: Size, modelName: string) {\n if (warned) return;\n if (modelName.toLowerCase().includes('gpt-4o')) {\n const warningMsg = `GPT-4o has a maximum image input size of 2000x768 or 768x2000, but got ${size.width}x${size.height}. Please set your interface to a smaller resolution. Otherwise, the result may be inaccurate.`;\n\n if (\n Math.max(size.width, size.height) > 2000 ||\n Math.min(size.width, size.height) > 768\n ) {\n console.warn(warningMsg);\n warned = true;\n }\n } else if (size.width > 1800 || size.height > 1800) {\n console.warn(\n `The image size seems too large (${size.width}x${size.height}). It may lead to more token usage, slower response, and inaccurate result.`,\n );\n warned = true;\n }\n}\n\nexport function mergeRects(rects: Rect[]) {\n const minLeft = Math.min(...rects.map((r) => r.left));\n const minTop = Math.min(...rects.map((r) => r.top));\n const maxRight = Math.max(...rects.map((r) => r.left + r.width));\n const maxBottom = Math.max(...rects.map((r) => r.top + r.height));\n return {\n left: minLeft,\n top: minTop,\n width: maxRight - minLeft,\n height: maxBottom - minTop,\n };\n}\n\n// expand the search area to at least 300 x 300, or add a default padding\nexport function expandSearchArea(\n rect: Rect,\n screenSize: Size,\n vlMode: TVlModeTypes | undefined,\n) {\n const minEdgeSize = vlMode === 'doubao-vision' ? 500 : 300;\n const defaultPadding = 160;\n\n const paddingSizeHorizontal =\n rect.width < minEdgeSize\n ? Math.ceil((minEdgeSize - rect.width) / 2)\n : defaultPadding;\n const paddingSizeVertical =\n rect.height < minEdgeSize\n ? Math.ceil((minEdgeSize - rect.height) / 2)\n : defaultPadding;\n rect.left = Math.max(0, rect.left - paddingSizeHorizontal);\n rect.width = Math.min(\n rect.width + paddingSizeHorizontal * 2,\n screenSize.width - rect.left,\n );\n rect.top = Math.max(0, rect.top - paddingSizeVertical);\n rect.height = Math.min(\n rect.height + paddingSizeVertical * 2,\n screenSize.height - rect.top,\n );\n return rect;\n}\n\nexport async function markupImageForLLM(\n screenshotBase64: string,\n tree: ElementTreeNode<BaseElement>,\n size: Size,\n) {\n const elementsInfo = treeToList(tree);\n const elementsPositionInfoWithoutText = elementsInfo!.filter(\n (elementInfo) => {\n if (elementInfo.attributes.nodeType === NodeType.TEXT) {\n return false;\n }\n return true;\n },\n );\n\n const imagePayload = await compositeElementInfoImg({\n inputImgBase64: screenshotBase64,\n elementsPositionInfo: elementsPositionInfoWithoutText,\n size,\n });\n return imagePayload;\n}\n\nexport function buildYamlFlowFromPlans(\n plans: PlanningAction[],\n actionSpace: DeviceAction<any>[],\n sleep?: number,\n): MidsceneYamlFlowItem[] {\n const flow: MidsceneYamlFlowItem[] = [];\n\n for (const plan of plans) {\n const verb = plan.type;\n\n const action = actionSpace.find((action) => action.name === verb);\n if (!action) {\n console.warn(\n `Cannot convert action ${verb} to yaml flow. Will ignore it.`,\n );\n continue;\n }\n\n const flowKey = action.interfaceAlias || verb;\n const flowParam = action.paramSchema\n ? dumpActionParam(plan.param || {}, action.paramSchema)\n : {};\n\n const flowItem: MidsceneYamlFlowItem = {\n [flowKey]: '',\n ...flowParam,\n };\n\n flow.push(flowItem);\n }\n\n if (sleep) {\n flow.push({\n sleep,\n });\n }\n\n return flow;\n}\n\n// Zod schemas for shared types\nexport const PointSchema = z.object({\n left: z.number(),\n top: z.number(),\n});\n\nexport const SizeSchema = z.object({\n width: z.number(),\n height: z.number(),\n dpr: z.number().optional(),\n});\n\nexport const RectSchema = PointSchema.and(SizeSchema).and(\n z.object({\n zoom: z.number().optional(),\n }),\n);\n\n// Zod schema for TMultimodalPrompt\nexport const TMultimodalPromptSchema = z.object({\n images: z\n .array(\n z.object({\n name: z.string(),\n url: z.string(),\n }),\n )\n .optional(),\n convertHttpImage2Base64: z.boolean().optional(),\n});\n\n// Zod schema for TUserPrompt\nexport const TUserPromptSchema = z.union([\n z.string(),\n z\n .object({\n prompt: z.string(),\n })\n .and(TMultimodalPromptSchema.partial()),\n]);\n\n// Generate TypeScript types from Zod schemas\nexport type TMultimodalPrompt = z.infer<typeof TMultimodalPromptSchema>;\nexport type TUserPrompt = z.infer<typeof TUserPromptSchema>;\n\nconst locateFieldFlagName = 'midscene_location_field_flag';\n\nconst MidsceneLocationResult = z\n .object({\n [locateFieldFlagName]: z.literal(true),\n prompt: TUserPromptSchema,\n\n // optional fields\n deepThink: z.boolean().optional(), // only available in vl model\n cacheable: z.boolean().optional(),\n xpath: z.boolean().optional(), // preset result for xpath\n\n // these two fields will only appear in the result\n center: z.tuple([z.number(), z.number()]),\n rect: RectSchema,\n })\n .passthrough();\n\nexport type MidsceneLocationResultType = z.infer<typeof MidsceneLocationResult>;\nexport const getMidsceneLocationSchema = () => {\n return MidsceneLocationResult;\n};\n\nexport const ifMidsceneLocatorField = (field: any): boolean => {\n // Handle optional fields by getting the inner type\n let actualField = field;\n if (actualField._def?.typeName === 'ZodOptional') {\n actualField = actualField._def.innerType;\n }\n\n // Check if this is a ZodUnion (the new MidsceneLocation structure)\n if (actualField._def?.typeName === 'ZodObject') {\n const shape = actualField._def.shape();\n return locateFieldFlagName in shape;\n }\n\n return false;\n};\n\nexport const dumpMidsceneLocatorField = (field: any): string => {\n assert(\n ifMidsceneLocatorField(field),\n 'field is not a midscene locator field',\n );\n\n // If field is a string, return it directly\n if (typeof field === 'string') {\n return field;\n }\n\n // If field is an object with prompt property\n if (field && typeof field === 'object' && field.prompt) {\n // If prompt is a string, return it directly\n if (typeof field.prompt === 'string') {\n return field.prompt;\n }\n // If prompt is a TUserPrompt object, extract the prompt string\n if (typeof field.prompt === 'object' && field.prompt.prompt) {\n return field.prompt.prompt; // TODO: dump images if necessary\n }\n }\n\n // Fallback: try to convert to string\n return String(field);\n};\n\nexport const findAllMidsceneLocatorField = (\n zodType?: z.ZodType<any>,\n requiredOnly?: boolean,\n): string[] => {\n if (!zodType) {\n return [];\n }\n\n // Check if this is a ZodObject by checking if it has a shape property\n const zodObject = zodType as any;\n if (zodObject._def?.typeName === 'ZodObject' && zodObject.shape) {\n const keys = Object.keys(zodObject.shape);\n return keys.filter((key) => {\n const field = zodObject.shape[key];\n if (!ifMidsceneLocatorField(field)) {\n return false;\n }\n\n // If requiredOnly is true, filter out optional fields\n if (requiredOnly) {\n return field._def?.typeName !== 'ZodOptional';\n }\n\n return true;\n });\n }\n\n // For other ZodType instances, we can't extract field names\n return [];\n};\n\nexport const dumpActionParam = (\n jsonObject: Record<string, any>,\n zodSchema: z.ZodType<any>,\n): Record<string, any> => {\n const locatorFields = findAllMidsceneLocatorField(zodSchema);\n const result = { ...jsonObject };\n\n for (const fieldName of locatorFields) {\n const fieldValue = result[fieldName];\n if (fieldValue) {\n // If it's already a string, keep it as is\n if (typeof fieldValue === 'string') {\n result[fieldName] = fieldValue;\n } else if (typeof fieldValue === 'object') {\n // Check if this field is actually a MidsceneLocationType object\n if (fieldValue.prompt) {\n // If prompt is a string, use it directly\n if (typeof fieldValue.prompt === 'string') {\n result[fieldName] = fieldValue.prompt;\n } else if (\n typeof fieldValue.prompt === 'object' &&\n fieldValue.prompt.prompt\n ) {\n // If prompt is a TUserPrompt object, extract the prompt string\n result[fieldName] = fieldValue.prompt.prompt;\n }\n }\n }\n }\n }\n\n return result;\n};\n\nexport const loadActionParam = (\n jsonObject: Record<string, any>,\n zodSchema: z.ZodType<any>,\n): Record<string, any> => {\n const locatorFields = findAllMidsceneLocatorField(zodSchema);\n const result = { ...jsonObject };\n\n for (const fieldName of locatorFields) {\n const fieldValue = result[fieldName];\n if (fieldValue && typeof fieldValue === 'string') {\n result[fieldName] = {\n [locateFieldFlagName]: true,\n prompt: fieldValue,\n };\n }\n }\n\n return result;\n};\n"],"names":["__webpack_require__","definition","key","Object","obj","prop","Symbol","AIActionType","defaultBboxSize","debugInspectUtils","getDebug","fillBboxParam","locate","width","height","vlMode","adaptBbox","adaptQwenBbox","bbox","msg","JSON","Error","result","Math","adaptDoubaoBbox","assert","splitted","Number","Array","bboxList","item","x","y","adaptGeminiBbox","left","top","right","bottom","adaptBboxToRect","offsetX","offsetY","rectLeft","rectTop","rectWidth","rectHeight","rect","warned","warnGPT4oSizeLimit","size","modelName","warningMsg","console","mergeRects","rects","minLeft","r","minTop","maxRight","maxBottom","expandSearchArea","screenSize","minEdgeSize","defaultPadding","paddingSizeHorizontal","paddingSizeVertical","markupImageForLLM","screenshotBase64","tree","elementsInfo","treeToList","elementsPositionInfoWithoutText","elementInfo","NodeType","imagePayload","compositeElementInfoImg","buildYamlFlowFromPlans","plans","actionSpace","sleep","flow","plan","verb","action","flowKey","flowParam","dumpActionParam","flowItem","PointSchema","z","SizeSchema","RectSchema","TMultimodalPromptSchema","TUserPromptSchema","locateFieldFlagName","MidsceneLocationResult","getMidsceneLocationSchema","ifMidsceneLocatorField","field","_actualField__def","_actualField__def1","actualField","shape","dumpMidsceneLocatorField","String","findAllMidsceneLocatorField","zodType","requiredOnly","_zodObject__def","zodObject","keys","_field__def","jsonObject","zodSchema","locatorFields","fieldName","fieldValue","loadActionParam"],"mappings":";;;IAAAA,oBAAoB,CAAC,GAAG,CAAC,UAASC;QACjC,IAAI,IAAIC,OAAOD,WACR,IAAGD,oBAAoB,CAAC,CAACC,YAAYC,QAAQ,CAACF,oBAAoB,CAAC,CAAC,UAASE,MACzEC,OAAO,cAAc,CAAC,UAASD,KAAK;YAAE,YAAY;YAAM,KAAKD,UAAU,CAACC,IAAI;QAAC;IAGzF;;;ICNAF,oBAAoB,CAAC,GAAG,CAACI,KAAKC,OAAUF,OAAO,SAAS,CAAC,cAAc,CAAC,IAAI,CAACC,KAAKC;;;ICClFL,oBAAoB,CAAC,GAAG,CAAC;QACxB,IAAG,AAAkB,eAAlB,OAAOM,UAA0BA,OAAO,WAAW,EACrDH,OAAO,cAAc,CAAC,UAASG,OAAO,WAAW,EAAE;YAAE,OAAO;QAAS;QAEtEH,OAAO,cAAc,CAAC,UAAS,cAAc;YAAE,OAAO;QAAK;IAC5D;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;ACkBO,IAAKI,sBAAYA,WAAAA,GAAAA,SAAZA,YAAY;;;;;;WAAZA;;AAQZ,MAAMC,kBAAkB;AACxB,MAAMC,oBAAoBC,AAAAA,IAAAA,uBAAAA,QAAAA,AAAAA,EAAS;AAG5B,SAASC,cACdC,MAA2B,EAC3BC,KAAa,EACbC,MAAc,EACdC,MAAgC;IAGhC,IAAKH,OAAe,OAAO,IAAI,CAACA,CAAAA,QAAAA,SAAAA,KAAAA,IAAAA,OAAQ,IAAI,AAAD,GAAG;QAC5CA,OAAO,IAAI,GAAIA,OAAe,OAAO;QAErC,OAAQA,OAAe,OAAO;IAChC;IAEA,IAAIA,QAAAA,SAAAA,KAAAA,IAAAA,OAAQ,IAAI,EACdA,OAAO,IAAI,GAAGI,UAAUJ,OAAO,IAAI,EAAEC,OAAOC,QAAQC;IAGtD,OAAOH;AACT;AAEO,SAASK,cACdC,IAAc;IAEd,IAAIA,KAAK,MAAM,GAAG,GAAG;QACnB,MAAMC,MAAM,CAAC,oCAAoC,EAAEC,KAAK,SAAS,CAACF,MAAM,CAAC,CAAC;QAC1E,MAAM,IAAIG,MAAMF;IAClB;IAEA,MAAMG,SAA2C;QAC/CC,KAAK,KAAK,CAACL,IAAI,CAAC,EAAE;QAClBK,KAAK,KAAK,CAACL,IAAI,CAAC,EAAE;QACC,YAAnB,OAAOA,IAAI,CAAC,EAAE,GACVK,KAAK,KAAK,CAACL,IAAI,CAAC,EAAE,IAClBK,KAAK,KAAK,CAACL,IAAI,CAAC,EAAE,GAAGV;QACN,YAAnB,OAAOU,IAAI,CAAC,EAAE,GACVK,KAAK,KAAK,CAACL,IAAI,CAAC,EAAE,IAClBK,KAAK,KAAK,CAACL,IAAI,CAAC,EAAE,GAAGV;KAC1B;IACD,OAAOc;AACT;AAEO,SAASE,gBACdN,IAAkC,EAClCL,KAAa,EACbC,MAAc;IAEdW,IAAAA,sBAAAA,MAAAA,AAAAA,EACEZ,QAAQ,KAAKC,SAAS,GACtB;IAGF,IAAI,AAAgB,YAAhB,OAAOI,MAAmB;QAC5BO,IAAAA,sBAAAA,MAAAA,AAAAA,EACE,+BAA+B,IAAI,CAACP,KAAK,IAAI,KAC7C,CAAC,iDAAiD,EAAEA,MAAM;QAE5D,MAAMQ,WAAWR,KAAK,KAAK,CAAC;QAC5B,IAAIQ,AAAoB,MAApBA,SAAS,MAAM,EACjB,OAAO;YACLH,KAAK,KAAK,CAAEI,OAAOD,QAAQ,CAAC,EAAE,IAAIb,QAAS;YAC3CU,KAAK,KAAK,CAAEI,OAAOD,QAAQ,CAAC,EAAE,IAAIZ,SAAU;YAC5CS,KAAK,KAAK,CAAEI,OAAOD,QAAQ,CAAC,EAAE,IAAIb,QAAS;YAC3CU,KAAK,KAAK,CAAEI,OAAOD,QAAQ,CAAC,EAAE,IAAIZ,SAAU;SAC7C;QAEH,MAAM,IAAIO,MAAM,CAAC,iDAAiD,EAAEH,MAAM;IAC5E;IAEA,IAAIU,MAAM,OAAO,CAACV,SAASU,MAAM,OAAO,CAACV,IAAI,CAAC,EAAE,GAC9CA,OAAOA,IAAI,CAAC,EAAE;IAGhB,IAAIW,WAAqB,EAAE;IAC3B,IAAID,MAAM,OAAO,CAACV,SAAS,AAAmB,YAAnB,OAAOA,IAAI,CAAC,EAAE,EACvCA,KAAK,OAAO,CAAC,CAACY;QACZ,IAAI,AAAgB,YAAhB,OAAOA,QAAqBA,KAAK,QAAQ,CAAC,MAAM;YAClD,MAAM,CAACC,GAAGC,EAAE,GAAGF,KAAK,KAAK,CAAC;YAC1BD,SAAS,IAAI,CAACF,OAAOI,EAAE,IAAI,KAAKJ,OAAOK,EAAE,IAAI;QAC/C,OAAO,IAAI,AAAgB,YAAhB,OAAOF,QAAqBA,KAAK,QAAQ,CAAC,MAAM;YACzD,MAAM,CAACC,GAAGC,EAAE,GAAGF,KAAK,KAAK,CAAC;YAC1BD,SAAS,IAAI,CAACF,OAAOI,EAAE,IAAI,KAAKJ,OAAOK,EAAE,IAAI;QAC/C,OACEH,SAAS,IAAI,CAACF,OAAOG;IAEzB;SAEAD,WAAWX;IAGb,IAAIW,AAAoB,MAApBA,SAAS,MAAM,IAAUA,AAAoB,MAApBA,SAAS,MAAM,EAC1C,OAAO;QACLN,KAAK,KAAK,CAAEM,QAAQ,CAAC,EAAE,GAAGhB,QAAS;QACnCU,KAAK,KAAK,CAAEM,QAAQ,CAAC,EAAE,GAAGf,SAAU;QACpCS,KAAK,KAAK,CAAEM,QAAQ,CAAC,EAAE,GAAGhB,QAAS;QACnCU,KAAK,KAAK,CAAEM,QAAQ,CAAC,EAAE,GAAGf,SAAU;KACrC;IAIH,IACEe,AAAoB,MAApBA,SAAS,MAAM,IACfA,AAAoB,MAApBA,SAAS,MAAM,IACfA,AAAoB,MAApBA,SAAS,MAAM,IACfA,AAAoB,MAApBA,SAAS,MAAM,EAEf,OAAO;QACLN,KAAK,GAAG,CACN,GACAA,KAAK,KAAK,CAAEM,QAAQ,CAAC,EAAE,GAAGhB,QAAS,QAAQL,kBAAkB;QAE/De,KAAK,GAAG,CACN,GACAA,KAAK,KAAK,CAAEM,QAAQ,CAAC,EAAE,GAAGf,SAAU,QAAQN,kBAAkB;QAEhEe,KAAK,GAAG,CACNV,OACAU,KAAK,KAAK,CAAEM,QAAQ,CAAC,EAAE,GAAGhB,QAAS,QAAQL,kBAAkB;QAE/De,KAAK,GAAG,CACNT,QACAS,KAAK,KAAK,CAAEM,QAAQ,CAAC,EAAE,GAAGf,SAAU,QAAQN,kBAAkB;KAEjE;IAGH,IAAIU,AAAgB,MAAhBA,KAAK,MAAM,EACb,OAAO;QACLK,KAAK,KAAK,CAAEM,QAAQ,CAAC,EAAE,GAAGhB,QAAS;QACnCU,KAAK,KAAK,CAAEM,QAAQ,CAAC,EAAE,GAAGf,SAAU;QACpCS,KAAK,KAAK,CAAEM,QAAQ,CAAC,EAAE,GAAGhB,QAAS;QACnCU,KAAK,KAAK,CAAEM,QAAQ,CAAC,EAAE,GAAGf,SAAU;KACrC;IAGH,MAAMK,MAAM,CAAC,0CAA0C,EAAEC,KAAK,SAAS,CAACF,MAAM,CAAC,CAAC;IAChF,MAAM,IAAIG,MAAMF;AAClB;AAEO,SAASH,UACdE,IAAc,EACdL,KAAa,EACbC,MAAc,EACdC,MAAgC;IAEhC,IAAIA,AAAW,oBAAXA,UAA8BA,AAAW,kBAAXA,QAChC,OAAOS,gBAAgBN,MAAML,OAAOC;IAGtC,IAAIC,AAAW,aAAXA,QACF,OAAOkB,gBAAgBf,MAAML,OAAOC;IAGtC,OAAOG,cAAcC;AACvB;AAEO,SAASe,gBACdf,IAAc,EACdL,KAAa,EACbC,MAAc;IAEd,MAAMoB,OAAOX,KAAK,KAAK,CAAEL,IAAI,CAAC,EAAE,GAAGL,QAAS;IAC5C,MAAMsB,MAAMZ,KAAK,KAAK,CAAEL,IAAI,CAAC,EAAE,GAAGJ,SAAU;IAC5C,MAAMsB,QAAQb,KAAK,KAAK,CAAEL,IAAI,CAAC,EAAE,GAAGL,QAAS;IAC7C,MAAMwB,SAASd,KAAK,KAAK,CAAEL,IAAI,CAAC,EAAE,GAAGJ,SAAU;IAC/C,OAAO;QAACoB;QAAMC;QAAKC;QAAOC;KAAO;AACnC;AAEO,SAASC,gBACdpB,IAAc,EACdL,KAAa,EACbC,MAAc,EACdyB,UAAU,CAAC,EACXC,UAAU,CAAC,EACXzB,MAAiC;IAEjCN,kBAAkB,mBAAmBS,MAAML,OAAOC,QAAQyB,SAASC;IACnE,MAAM,CAACN,MAAMC,KAAKC,OAAOC,OAAO,GAAGrB,UAAUE,MAAML,OAAOC,QAAQC;IAGlE,MAAM0B,WAAWP;IACjB,MAAMQ,UAAUP;IAChB,IAAIQ,YAAYP,QAAQF;IACxB,IAAIU,aAAaP,SAASF;IAI1B,IAAIM,WAAWE,YAAY9B,OACzB8B,YAAY9B,QAAQ4B;IAItB,IAAIC,UAAUE,aAAa9B,QACzB8B,aAAa9B,SAAS4B;IAIxBC,YAAYpB,KAAK,GAAG,CAAC,GAAGoB;IACxBC,aAAarB,KAAK,GAAG,CAAC,GAAGqB;IAEzB,MAAMC,OAAO;QACX,MAAMJ,WAAWF;QACjB,KAAKG,UAAUF;QACf,OAAOG;QACP,QAAQC;IACV;IACAnC,kBAAkB,4BAA4BoC;IAC9C,OAAOA;AACT;AAEA,IAAIC,SAAS;AACN,SAASC,mBAAmBC,IAAU,EAAEC,SAAiB;IAC9D,IAAIH,QAAQ;IACZ,IAAIG,UAAU,WAAW,GAAG,QAAQ,CAAC,WAAW;QAC9C,MAAMC,aAAa,CAAC,uEAAuE,EAAEF,KAAK,KAAK,CAAC,CAAC,EAAEA,KAAK,MAAM,CAAC,6FAA6F,CAAC;QAErN,IACEzB,KAAK,GAAG,CAACyB,KAAK,KAAK,EAAEA,KAAK,MAAM,IAAI,QACpCzB,KAAK,GAAG,CAACyB,KAAK,KAAK,EAAEA,KAAK,MAAM,IAAI,KACpC;YACAG,QAAQ,IAAI,CAACD;YACbJ,SAAS;QACX;IACF,OAAO,IAAIE,KAAK,KAAK,GAAG,QAAQA,KAAK,MAAM,GAAG,MAAM;QAClDG,QAAQ,IAAI,CACV,CAAC,gCAAgC,EAAEH,KAAK,KAAK,CAAC,CAAC,EAAEA,KAAK,MAAM,CAAC,2EAA2E,CAAC;QAE3IF,SAAS;IACX;AACF;AAEO,SAASM,WAAWC,KAAa;IACtC,MAAMC,UAAU/B,KAAK,GAAG,IAAI8B,MAAM,GAAG,CAAC,CAACE,IAAMA,EAAE,IAAI;IACnD,MAAMC,SAASjC,KAAK,GAAG,IAAI8B,MAAM,GAAG,CAAC,CAACE,IAAMA,EAAE,GAAG;IACjD,MAAME,WAAWlC,KAAK,GAAG,IAAI8B,MAAM,GAAG,CAAC,CAACE,IAAMA,EAAE,IAAI,GAAGA,EAAE,KAAK;IAC9D,MAAMG,YAAYnC,KAAK,GAAG,IAAI8B,MAAM,GAAG,CAAC,CAACE,IAAMA,EAAE,GAAG,GAAGA,EAAE,MAAM;IAC/D,OAAO;QACL,MAAMD;QACN,KAAKE;QACL,OAAOC,WAAWH;QAClB,QAAQI,YAAYF;IACtB;AACF;AAGO,SAASG,iBACdd,IAAU,EACVe,UAAgB,EAChB7C,MAAgC;IAEhC,MAAM8C,cAAc9C,AAAW,oBAAXA,SAA6B,MAAM;IACvD,MAAM+C,iBAAiB;IAEvB,MAAMC,wBACJlB,KAAK,KAAK,GAAGgB,cACTtC,KAAK,IAAI,CAAEsC,AAAAA,CAAAA,cAAchB,KAAK,KAAI,IAAK,KACvCiB;IACN,MAAME,sBACJnB,KAAK,MAAM,GAAGgB,cACVtC,KAAK,IAAI,CAAEsC,AAAAA,CAAAA,cAAchB,KAAK,MAAK,IAAK,KACxCiB;IACNjB,KAAK,IAAI,GAAGtB,KAAK,GAAG,CAAC,GAAGsB,KAAK,IAAI,GAAGkB;IACpClB,KAAK,KAAK,GAAGtB,KAAK,GAAG,CACnBsB,KAAK,KAAK,GAAGkB,AAAwB,IAAxBA,uBACbH,WAAW,KAAK,GAAGf,KAAK,IAAI;IAE9BA,KAAK,GAAG,GAAGtB,KAAK,GAAG,CAAC,GAAGsB,KAAK,GAAG,GAAGmB;IAClCnB,KAAK,MAAM,GAAGtB,KAAK,GAAG,CACpBsB,KAAK,MAAM,GAAGmB,AAAsB,IAAtBA,qBACdJ,WAAW,MAAM,GAAGf,KAAK,GAAG;IAE9B,OAAOA;AACT;AAEO,eAAeoB,kBACpBC,gBAAwB,EACxBC,IAAkC,EAClCnB,IAAU;IAEV,MAAMoB,eAAeC,AAAAA,IAAAA,0BAAAA,UAAAA,AAAAA,EAAWF;IAChC,MAAMG,kCAAkCF,aAAc,MAAM,CAC1D,CAACG;QACC,IAAIA,YAAY,UAAU,CAAC,QAAQ,KAAKC,0BAAAA,QAAAA,CAAAA,IAAa,EACnD,OAAO;QAET,OAAO;IACT;IAGF,MAAMC,eAAe,MAAMC,AAAAA,IAAAA,oBAAAA,uBAAAA,AAAAA,EAAwB;QACjD,gBAAgBR;QAChB,sBAAsBI;QACtBtB;IACF;IACA,OAAOyB;AACT;AAEO,SAASE,uBACdC,KAAuB,EACvBC,WAAgC,EAChCC,KAAc;IAEd,MAAMC,OAA+B,EAAE;IAEvC,KAAK,MAAMC,QAAQJ,MAAO;QACxB,MAAMK,OAAOD,KAAK,IAAI;QAEtB,MAAME,SAASL,YAAY,IAAI,CAAC,CAACK,SAAWA,OAAO,IAAI,KAAKD;QAC5D,IAAI,CAACC,QAAQ;YACX/B,QAAQ,IAAI,CACV,CAAC,sBAAsB,EAAE8B,KAAK,8BAA8B,CAAC;YAE/D;QACF;QAEA,MAAME,UAAUD,OAAO,cAAc,IAAID;QACzC,MAAMG,YAAYF,OAAO,WAAW,GAChCG,gBAAgBL,KAAK,KAAK,IAAI,CAAC,GAAGE,OAAO,WAAW,IACpD,CAAC;QAEL,MAAMI,WAAiC;YACrC,CAACH,QAAQ,EAAE;YACX,GAAGC,SAAS;QACd;QAEAL,KAAK,IAAI,CAACO;IACZ;IAEA,IAAIR,OACFC,KAAK,IAAI,CAAC;QACRD;IACF;IAGF,OAAOC;AACT;AAGO,MAAMQ,cAAcC,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IAClC,MAAMA,6BAAAA,CAAAA,CAAAA,MAAQ;IACd,KAAKA,6BAAAA,CAAAA,CAAAA,MAAQ;AACf;AAEO,MAAMC,aAAaD,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IACjC,OAAOA,6BAAAA,CAAAA,CAAAA,MAAQ;IACf,QAAQA,6BAAAA,CAAAA,CAAAA,MAAQ;IAChB,KAAKA,6BAAAA,CAAAA,CAAAA,MAAQ,GAAG,QAAQ;AAC1B;AAEO,MAAME,aAAaH,YAAY,GAAG,CAACE,YAAY,GAAG,CACvDD,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IACP,MAAMA,6BAAAA,CAAAA,CAAAA,MAAQ,GAAG,QAAQ;AAC3B;AAIK,MAAMG,0BAA0BH,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IAC9C,QAAQA,6BAAAA,CAAAA,CAAAA,KACA,CACJA,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;QACP,MAAMA,6BAAAA,CAAAA,CAAAA,MAAQ;QACd,KAAKA,6BAAAA,CAAAA,CAAAA,MAAQ;IACf,IAED,QAAQ;IACX,yBAAyBA,6BAAAA,CAAAA,CAAAA,OAAS,GAAG,QAAQ;AAC/C;AAGO,MAAMI,oBAAoBJ,6BAAAA,CAAAA,CAAAA,KAAO,CAAC;IACvCA,6BAAAA,CAAAA,CAAAA,MAAQ;IACRA,6BAAAA,CAAAA,CAAAA,MACS,CAAC;QACN,QAAQA,6BAAAA,CAAAA,CAAAA,MAAQ;IAClB,GACC,GAAG,CAACG,wBAAwB,OAAO;CACvC;AAMD,MAAME,sBAAsB;AAE5B,MAAMC,yBAAyBN,6BAAAA,CAAAA,CAAAA,MACtB,CAAC;IACN,CAACK,oBAAoB,EAAEL,6BAAAA,CAAAA,CAAAA,OAAS,CAAC;IACjC,QAAQI;IAGR,WAAWJ,6BAAAA,CAAAA,CAAAA,OAAS,GAAG,QAAQ;IAC/B,WAAWA,6BAAAA,CAAAA,CAAAA,OAAS,GAAG,QAAQ;IAC/B,OAAOA,6BAAAA,CAAAA,CAAAA,OAAS,GAAG,QAAQ;IAG3B,QAAQA,6BAAAA,CAAAA,CAAAA,KAAO,CAAC;QAACA,6BAAAA,CAAAA,CAAAA,MAAQ;QAAIA,6BAAAA,CAAAA,CAAAA,MAAQ;KAAG;IACxC,MAAME;AACR,GACC,WAAW;AAGP,MAAMK,4BAA4B,IAChCD;AAGF,MAAME,yBAAyB,CAACC;QAGjCC,mBAKAC;IANJ,IAAIC,cAAcH;IAClB,IAAIC,AAAAA,SAAAA,CAAAA,oBAAAA,YAAY,IAAI,AAAD,IAAfA,KAAAA,IAAAA,kBAAkB,QAAQ,AAAD,MAAM,eACjCE,cAAcA,YAAY,IAAI,CAAC,SAAS;IAI1C,IAAID,AAAAA,SAAAA,CAAAA,qBAAAA,YAAY,IAAI,AAAD,IAAfA,KAAAA,IAAAA,mBAAkB,QAAQ,AAAD,MAAM,aAAa;QAC9C,MAAME,QAAQD,YAAY,IAAI,CAAC,KAAK;QACpC,OAAOP,uBAAuBQ;IAChC;IAEA,OAAO;AACT;AAEO,MAAMC,2BAA2B,CAACL;IACvCxE,IAAAA,sBAAAA,MAAAA,AAAAA,EACEuE,uBAAuBC,QACvB;IAIF,IAAI,AAAiB,YAAjB,OAAOA,OACT,OAAOA;IAIT,IAAIA,SAAS,AAAiB,YAAjB,OAAOA,SAAsBA,MAAM,MAAM,EAAE;QAEtD,IAAI,AAAwB,YAAxB,OAAOA,MAAM,MAAM,EACrB,OAAOA,MAAM,MAAM;QAGrB,IAAI,AAAwB,YAAxB,OAAOA,MAAM,MAAM,IAAiBA,MAAM,MAAM,CAAC,MAAM,EACzD,OAAOA,MAAM,MAAM,CAAC,MAAM;IAE9B;IAGA,OAAOM,OAAON;AAChB;AAEO,MAAMO,8BAA8B,CACzCC,SACAC;QAQIC;IANJ,IAAI,CAACF,SACH,OAAO,EAAE;IAIX,MAAMG,YAAYH;IAClB,IAAIE,AAAAA,SAAAA,CAAAA,kBAAAA,UAAU,IAAI,AAAD,IAAbA,KAAAA,IAAAA,gBAAgB,QAAQ,AAAD,MAAM,eAAeC,UAAU,KAAK,EAAE;QAC/D,MAAMC,OAAO1G,OAAO,IAAI,CAACyG,UAAU,KAAK;QACxC,OAAOC,KAAK,MAAM,CAAC,CAAC3G;YAClB,MAAM+F,QAAQW,UAAU,KAAK,CAAC1G,IAAI;YAClC,IAAI,CAAC8F,uBAAuBC,QAC1B,OAAO;YAIT,IAAIS,cAAc;oBACTI;gBAAP,OAAOA,AAAAA,SAAAA,CAAAA,cAAAA,MAAM,IAAI,AAAD,IAATA,KAAAA,IAAAA,YAAY,QAAQ,AAAD,MAAM;YAClC;YAEA,OAAO;QACT;IACF;IAGA,OAAO,EAAE;AACX;AAEO,MAAMzB,kBAAkB,CAC7B0B,YACAC;IAEA,MAAMC,gBAAgBT,4BAA4BQ;IAClD,MAAM1F,SAAS;QAAE,GAAGyF,UAAU;IAAC;IAE/B,KAAK,MAAMG,aAAaD,cAAe;QACrC,MAAME,aAAa7F,MAAM,CAAC4F,UAAU;QACpC,IAAIC,YAEF;YAAA,IAAI,AAAsB,YAAtB,OAAOA,YACT7F,MAAM,CAAC4F,UAAU,GAAGC;iBACf,IAAI,AAAsB,YAAtB,OAAOA,YAEhB;gBAAA,IAAIA,WAAW,MAAM,EAEnB;oBAAA,IAAI,AAA6B,YAA7B,OAAOA,WAAW,MAAM,EAC1B7F,MAAM,CAAC4F,UAAU,GAAGC,WAAW,MAAM;yBAChC,IACL,AAA6B,YAA7B,OAAOA,WAAW,MAAM,IACxBA,WAAW,MAAM,CAAC,MAAM,EAGxB7F,MAAM,CAAC4F,UAAU,GAAGC,WAAW,MAAM,CAAC,MAAM;gBAC9C;YACF;QACF;IAEJ;IAEA,OAAO7F;AACT;AAEO,MAAM8F,kBAAkB,CAC7BL,YACAC;IAEA,MAAMC,gBAAgBT,4BAA4BQ;IAClD,MAAM1F,SAAS;QAAE,GAAGyF,UAAU;IAAC;IAE/B,KAAK,MAAMG,aAAaD,cAAe;QACrC,MAAME,aAAa7F,MAAM,CAAC4F,UAAU;QACpC,IAAIC,cAAc,AAAsB,YAAtB,OAAOA,YACvB7F,MAAM,CAAC4F,UAAU,GAAG;YAClB,CAACrB,oBAAoB,EAAE;YACvB,QAAQsB;QACV;IAEJ;IAEA,OAAO7F;AACT"}
@@ -26,30 +26,29 @@ __webpack_require__.r(__webpack_exports__);
26
26
  __webpack_require__.d(__webpack_exports__, {
27
27
  plan: ()=>external_llm_planning_js_namespaceObject.plan,
28
28
  SizeSchema: ()=>external_common_js_namespaceObject.SizeSchema,
29
- callAiFnWithStringResponse: ()=>index_js_namespaceObject.callAiFnWithStringResponse,
29
+ generateYamlTest: ()=>yaml_generator_js_namespaceObject.generateYamlTest,
30
30
  TMultimodalPromptSchema: ()=>external_common_js_namespaceObject.TMultimodalPromptSchema,
31
31
  elementByPositionWithElementInfo: ()=>util_js_namespaceObject.elementByPositionWithElementInfo,
32
32
  AiLocateSection: ()=>external_inspect_js_namespaceObject.AiLocateSection,
33
33
  AIActionType: ()=>external_common_js_namespaceObject.AIActionType,
34
34
  TUserPromptSchema: ()=>external_common_js_namespaceObject.TUserPromptSchema,
35
- callAi: ()=>index_js_namespaceObject.call,
35
+ callAI: ()=>index_js_namespaceObject.callAI,
36
36
  findAllMidsceneLocatorField: ()=>external_common_js_namespaceObject.findAllMidsceneLocatorField,
37
37
  AiExtractElementInfo: ()=>external_inspect_js_namespaceObject.AiExtractElementInfo,
38
38
  generatePlaywrightTest: ()=>playwright_generator_js_namespaceObject.generatePlaywrightTest,
39
- generateYamlTest: ()=>yaml_generator_js_namespaceObject.generateYamlTest,
40
39
  generateYamlTestStream: ()=>yaml_generator_js_namespaceObject.generateYamlTestStream,
41
40
  loadActionParam: ()=>external_common_js_namespaceObject.loadActionParam,
42
41
  resizeImageForUiTars: ()=>external_ui_tars_planning_js_namespaceObject.resizeImageForUiTars,
43
42
  systemPromptToLocateElement: ()=>llm_locator_js_namespaceObject.systemPromptToLocateElement,
43
+ vlmPlanning: ()=>external_ui_tars_planning_js_namespaceObject.vlmPlanning,
44
44
  RectSchema: ()=>external_common_js_namespaceObject.RectSchema,
45
45
  PointSchema: ()=>external_common_js_namespaceObject.PointSchema,
46
46
  adaptBboxToRect: ()=>external_common_js_namespaceObject.adaptBboxToRect,
47
- vlmPlanning: ()=>external_ui_tars_planning_js_namespaceObject.vlmPlanning,
47
+ callAIWithObjectResponse: ()=>index_js_namespaceObject.callAIWithObjectResponse,
48
48
  AiLocateElement: ()=>external_inspect_js_namespaceObject.AiLocateElement,
49
- callAiFn: ()=>external_common_js_namespaceObject.callAiFn,
50
- describeUserPage: ()=>util_js_namespaceObject.describeUserPage,
51
49
  getMidsceneLocationSchema: ()=>external_common_js_namespaceObject.getMidsceneLocationSchema,
52
- callToGetJSONObject: ()=>index_js_namespaceObject.callToGetJSONObject,
50
+ callAIWithStringResponse: ()=>index_js_namespaceObject.callAIWithStringResponse,
51
+ describeUserPage: ()=>util_js_namespaceObject.describeUserPage,
53
52
  generatePlaywrightTestStream: ()=>playwright_generator_js_namespaceObject.generatePlaywrightTestStream,
54
53
  dumpActionParam: ()=>external_common_js_namespaceObject.dumpActionParam
55
54
  });
@@ -72,10 +71,9 @@ exports.SizeSchema = __webpack_exports__.SizeSchema;
72
71
  exports.TMultimodalPromptSchema = __webpack_exports__.TMultimodalPromptSchema;
73
72
  exports.TUserPromptSchema = __webpack_exports__.TUserPromptSchema;
74
73
  exports.adaptBboxToRect = __webpack_exports__.adaptBboxToRect;
75
- exports.callAi = __webpack_exports__.callAi;
76
- exports.callAiFn = __webpack_exports__.callAiFn;
77
- exports.callAiFnWithStringResponse = __webpack_exports__.callAiFnWithStringResponse;
78
- exports.callToGetJSONObject = __webpack_exports__.callToGetJSONObject;
74
+ exports.callAI = __webpack_exports__.callAI;
75
+ exports.callAIWithObjectResponse = __webpack_exports__.callAIWithObjectResponse;
76
+ exports.callAIWithStringResponse = __webpack_exports__.callAIWithStringResponse;
79
77
  exports.describeUserPage = __webpack_exports__.describeUserPage;
80
78
  exports.dumpActionParam = __webpack_exports__.dumpActionParam;
81
79
  exports.elementByPositionWithElementInfo = __webpack_exports__.elementByPositionWithElementInfo;
@@ -101,10 +99,9 @@ for(var __webpack_i__ in __webpack_exports__)if (-1 === [
101
99
  "TMultimodalPromptSchema",
102
100
  "TUserPromptSchema",
103
101
  "adaptBboxToRect",
104
- "callAi",
105
- "callAiFn",
106
- "callAiFnWithStringResponse",
107
- "callToGetJSONObject",
102
+ "callAI",
103
+ "callAIWithObjectResponse",
104
+ "callAIWithStringResponse",
108
105
  "describeUserPage",
109
106
  "dumpActionParam",
110
107
  "elementByPositionWithElementInfo",
@@ -28,7 +28,6 @@ __webpack_require__.d(__webpack_exports__, {
28
28
  AiLocateSection: ()=>AiLocateSection,
29
29
  AiExtractElementInfo: ()=>AiExtractElementInfo
30
30
  });
31
- const env_namespaceObject = require("@midscene/shared/env");
32
31
  const img_namespaceObject = require("@midscene/shared/img");
33
32
  const logger_namespaceObject = require("@midscene/shared/logger");
34
33
  const utils_namespaceObject = require("@midscene/shared/utils");
@@ -85,25 +84,25 @@ const promptsToChatParam = async (multimodalPrompt)=>{
85
84
  return msgs;
86
85
  };
87
86
  async function AiLocateElement(options) {
88
- const { context, targetElementDescription, callAI } = options;
87
+ const { context, targetElementDescription, callAIFn, modelConfig } = options;
88
+ const { vlMode } = modelConfig;
89
89
  const { screenshotBase64 } = context;
90
- const modelPreferences = {
91
- intent: 'grounding'
92
- };
93
- const { description, elementById, insertElementByPosition } = await (0, util_js_namespaceObject.describeUserPage)(context, modelPreferences);
90
+ const { description, elementById, insertElementByPosition } = await (0, util_js_namespaceObject.describeUserPage)(context, {
91
+ vlMode
92
+ });
94
93
  (0, utils_namespaceObject.assert)(targetElementDescription, "cannot find the target element description");
95
94
  const userInstructionPrompt = await llm_locator_js_namespaceObject.findElementPrompt.format({
96
95
  pageDescription: description,
97
96
  targetElementDescription: extraTextFromUserPrompt(targetElementDescription)
98
97
  });
99
- const systemPrompt = (0, llm_locator_js_namespaceObject.systemPromptToLocateElement)((0, env_namespaceObject.vlLocateMode)(modelPreferences));
98
+ const systemPrompt = (0, llm_locator_js_namespaceObject.systemPromptToLocateElement)(vlMode);
100
99
  let imagePayload = screenshotBase64;
101
100
  if (options.searchConfig) {
102
101
  (0, utils_namespaceObject.assert)(options.searchConfig.rect, 'searchArea is provided but its rect cannot be found. Failed to locate element');
103
102
  (0, utils_namespaceObject.assert)(options.searchConfig.imageBase64, 'searchArea is provided but its imageBase64 cannot be found. Failed to locate element');
104
103
  imagePayload = options.searchConfig.imageBase64;
105
- } else if ('qwen-vl' === (0, env_namespaceObject.vlLocateMode)(modelPreferences)) imagePayload = await (0, img_namespaceObject.paddingToMatchBlockByBase64)(imagePayload);
106
- else if (!(0, env_namespaceObject.vlLocateMode)(modelPreferences)) imagePayload = await (0, external_common_js_namespaceObject.markupImageForLLM)(screenshotBase64, context.tree, context.size);
104
+ } else if ('qwen-vl' === vlMode) imagePayload = await (0, img_namespaceObject.paddingToMatchBlockByBase64)(imagePayload);
105
+ else if (!vlMode) imagePayload = await (0, external_common_js_namespaceObject.markupImageForLLM)(screenshotBase64, context.tree, context.size);
107
106
  const msgs = [
108
107
  {
109
108
  role: 'system',
@@ -133,10 +132,7 @@ async function AiLocateElement(options) {
133
132
  });
134
133
  msgs.push(...addOns);
135
134
  }
136
- const callAIFn = callAI || index_js_namespaceObject.callToGetJSONObject;
137
- const res = await callAIFn(msgs, external_common_js_namespaceObject.AIActionType.INSPECT_ELEMENT, {
138
- intent: 'grounding'
139
- });
135
+ const res = await callAIFn(msgs, external_common_js_namespaceObject.AIActionType.INSPECT_ELEMENT, modelConfig);
140
136
  const rawResponse = JSON.stringify(res.content);
141
137
  let resRect;
142
138
  let matchedElements = 'elements' in res.content ? res.content.elements : [];
@@ -144,7 +140,7 @@ async function AiLocateElement(options) {
144
140
  try {
145
141
  if ('bbox' in res.content && Array.isArray(res.content.bbox)) {
146
142
  var _options_searchConfig_rect, _options_searchConfig, _options_searchConfig_rect1, _options_searchConfig1, _options_searchConfig_rect2, _options_searchConfig2, _options_searchConfig_rect3, _options_searchConfig3;
147
- resRect = (0, external_common_js_namespaceObject.adaptBboxToRect)(res.content.bbox, (null == (_options_searchConfig = options.searchConfig) ? void 0 : null == (_options_searchConfig_rect = _options_searchConfig.rect) ? void 0 : _options_searchConfig_rect.width) || context.size.width, (null == (_options_searchConfig1 = options.searchConfig) ? void 0 : null == (_options_searchConfig_rect1 = _options_searchConfig1.rect) ? void 0 : _options_searchConfig_rect1.height) || context.size.height, modelPreferences, null == (_options_searchConfig2 = options.searchConfig) ? void 0 : null == (_options_searchConfig_rect2 = _options_searchConfig2.rect) ? void 0 : _options_searchConfig_rect2.left, null == (_options_searchConfig3 = options.searchConfig) ? void 0 : null == (_options_searchConfig_rect3 = _options_searchConfig3.rect) ? void 0 : _options_searchConfig_rect3.top);
143
+ resRect = (0, external_common_js_namespaceObject.adaptBboxToRect)(res.content.bbox, (null == (_options_searchConfig = options.searchConfig) ? void 0 : null == (_options_searchConfig_rect = _options_searchConfig.rect) ? void 0 : _options_searchConfig_rect.width) || context.size.width, (null == (_options_searchConfig1 = options.searchConfig) ? void 0 : null == (_options_searchConfig_rect1 = _options_searchConfig1.rect) ? void 0 : _options_searchConfig_rect1.height) || context.size.height, null == (_options_searchConfig2 = options.searchConfig) ? void 0 : null == (_options_searchConfig_rect2 = _options_searchConfig2.rect) ? void 0 : _options_searchConfig_rect2.left, null == (_options_searchConfig3 = options.searchConfig) ? void 0 : null == (_options_searchConfig_rect3 = _options_searchConfig3.rect) ? void 0 : _options_searchConfig_rect3.top, vlMode);
148
144
  debugInspect('resRect', resRect);
149
145
  const rectCenter = {
150
146
  x: resRect.left + resRect.width / 2,
@@ -183,12 +179,10 @@ async function AiLocateElement(options) {
183
179
  };
184
180
  }
185
181
  async function AiLocateSection(options) {
186
- const { context, sectionDescription } = options;
182
+ const { context, sectionDescription, modelConfig } = options;
183
+ const { vlMode } = modelConfig;
187
184
  const { screenshotBase64 } = context;
188
- const modelPreferences = {
189
- intent: 'grounding'
190
- };
191
- const systemPrompt = (0, llm_section_locator_js_namespaceObject.systemPromptToLocateSection)((0, env_namespaceObject.vlLocateMode)(modelPreferences));
185
+ const systemPrompt = (0, llm_section_locator_js_namespaceObject.systemPromptToLocateSection)(vlMode);
192
186
  const sectionLocatorInstructionText = await llm_section_locator_js_namespaceObject.sectionLocatorInstruction.format({
193
187
  sectionDescription: extraTextFromUserPrompt(sectionDescription)
194
188
  });
@@ -221,30 +215,26 @@ async function AiLocateSection(options) {
221
215
  });
222
216
  msgs.push(...addOns);
223
217
  }
224
- const result = await (0, external_common_js_namespaceObject.callAiFn)(msgs, external_common_js_namespaceObject.AIActionType.EXTRACT_DATA, {
225
- intent: 'grounding'
226
- });
218
+ const result = await (0, index_js_namespaceObject.callAIWithObjectResponse)(msgs, external_common_js_namespaceObject.AIActionType.EXTRACT_DATA, modelConfig);
227
219
  let sectionRect;
228
220
  const sectionBbox = result.content.bbox;
229
221
  if (sectionBbox) {
230
- const targetRect = (0, external_common_js_namespaceObject.adaptBboxToRect)(sectionBbox, context.size.width, context.size.height, modelPreferences);
222
+ const targetRect = (0, external_common_js_namespaceObject.adaptBboxToRect)(sectionBbox, context.size.width, context.size.height, 0, 0, vlMode);
231
223
  debugSection('original targetRect %j', targetRect);
232
224
  const referenceBboxList = result.content.references_bbox || [];
233
225
  debugSection('referenceBboxList %j', referenceBboxList);
234
- const referenceRects = referenceBboxList.filter((bbox)=>Array.isArray(bbox)).map((bbox)=>(0, external_common_js_namespaceObject.adaptBboxToRect)(bbox, context.size.width, context.size.height, modelPreferences));
226
+ const referenceRects = referenceBboxList.filter((bbox)=>Array.isArray(bbox)).map((bbox)=>(0, external_common_js_namespaceObject.adaptBboxToRect)(bbox, context.size.width, context.size.height, 0, 0, vlMode));
235
227
  debugSection('referenceRects %j', referenceRects);
236
228
  const mergedRect = (0, external_common_js_namespaceObject.mergeRects)([
237
229
  targetRect,
238
230
  ...referenceRects
239
231
  ]);
240
232
  debugSection('mergedRect %j', mergedRect);
241
- sectionRect = (0, external_common_js_namespaceObject.expandSearchArea)(mergedRect, context.size, modelPreferences);
233
+ sectionRect = (0, external_common_js_namespaceObject.expandSearchArea)(mergedRect, context.size, vlMode);
242
234
  debugSection('expanded sectionRect %j', sectionRect);
243
235
  }
244
236
  let imageBase64 = screenshotBase64;
245
- if (sectionRect) imageBase64 = await (0, img_namespaceObject.cropByRect)(screenshotBase64, sectionRect, (0, env_namespaceObject.getIsUseQwenVl)({
246
- intent: 'grounding'
247
- }));
237
+ if (sectionRect) imageBase64 = await (0, img_namespaceObject.cropByRect)(screenshotBase64, sectionRect, 'qwen-vl' === vlMode);
248
238
  return {
249
239
  rect: sectionRect,
250
240
  imageBase64,
@@ -255,14 +245,16 @@ async function AiLocateSection(options) {
255
245
  }
256
246
  async function AiExtractElementInfo(options) {
257
247
  var _options_extractOption;
258
- const { dataQuery, context, extractOption, multimodalPrompt, modelPreferences } = options;
248
+ const { dataQuery, context, extractOption, multimodalPrompt, modelConfig } = options;
249
+ const { vlMode } = modelConfig;
259
250
  const systemPrompt = (0, extraction_js_namespaceObject.systemPromptToExtract)();
260
251
  const { screenshotBase64 } = context;
261
- const { description, elementById } = await (0, util_js_namespaceObject.describeUserPage)(context, modelPreferences, {
252
+ const { description, elementById } = await (0, util_js_namespaceObject.describeUserPage)(context, {
262
253
  truncateTextLength: 200,
263
254
  filterNonTextContent: false,
264
255
  visibleOnly: false,
265
- domIncluded: null == extractOption ? void 0 : extractOption.domIncluded
256
+ domIncluded: null == extractOption ? void 0 : extractOption.domIncluded,
257
+ vlMode
266
258
  });
267
259
  const extractDataPromptText = await (0, extraction_js_namespaceObject.extractDataQueryPrompt)(description, dataQuery);
268
260
  const userContent = [];
@@ -298,7 +290,7 @@ async function AiExtractElementInfo(options) {
298
290
  });
299
291
  msgs.push(...addOns);
300
292
  }
301
- const result = await (0, external_common_js_namespaceObject.callAiFn)(msgs, external_common_js_namespaceObject.AIActionType.EXTRACT_DATA, modelPreferences);
293
+ const result = await (0, index_js_namespaceObject.callAIWithObjectResponse)(msgs, external_common_js_namespaceObject.AIActionType.EXTRACT_DATA, modelConfig);
302
294
  return {
303
295
  parseResult: result.content,
304
296
  elementById,