@midscene/core 1.0.4 → 1.0.5-beta-20251230124359.0

package/dist/lib/agent/utils.js

@@ -148,7 +148,7 @@ async function matchElementFromCache(context, cacheEntry, cachePrompt, cacheable
         return;
     }
 }
-const getMidsceneVersion = ()=>"1.0.4";
+const getMidsceneVersion = ()=>"1.0.5-beta-20251230124359.0";
 const parsePrompt = (prompt)=>{
     if ('string' == typeof prompt) return {
         textPrompt: prompt,

package/dist/lib/ai-model/prompt/llm-planning.js

@@ -77,7 +77,14 @@ async function systemPromptToTaskPlanning({ actionSpace, vlMode, includeBbox })
     if (includeBbox && !vlMode) throw new Error('vlMode cannot be undefined when includeBbox is true. A valid vlMode is required for bbox-based location.');
     const actionDescriptionList = actionSpace.map((action)=>descriptionForAction(action, vlLocateParam(includeBbox ? vlMode : void 0)));
     const actionList = actionDescriptionList.join('\n');
-    const logFieldInstruction = `
+    const logFieldInstruction = 'qwen2.5-vl' === vlMode || 'qwen3-vl' === vlMode ? `
+## About the \`log\` field
+
+* Log your thoughts and what the next one action (ONLY ONE!) you can do according to the screenshot and the instruction. 
+* The log should contain the following information: "The user wants to do ... . According to the instruction and the previous logs, next step is to .... Now i am going to compose an action '{ action-type }' to do ....". 
+* If no action should be done, log the reason. 
+* Use the same language as the user's instruction.
+` : `
 ## About the \`log\` field (preamble message)
 
 The \`log\` field is a brief preamble message to the user explaining what you’re about to do. It should follow these principles and examples:
@@ -93,6 +100,7 @@ The \`log\` field is a brief preamble message to the user explaining what you’
 - "Previous actions failed to find the 'Yes' button, i will try again"
 - "Go back to find the login button"
 `;
+    const logSample = 'qwen2.5-vl' === vlMode || 'qwen3-vl' === vlMode ? 'The user wants to fill the form and login. According to the instruction and the previous logs, next step is to click the login button. Now i am going to compose an action "Tap" to click the login button.' : 'Click the login button';
     return `
 Target: User will give you an instruction, some screenshots and previous logs indicating what have been done. Your task is to plan the next one action according to current situation to accomplish the instruction.
 
@@ -117,7 +125,7 @@ ${logFieldInstruction}
 
 Return in JSON format:
 {
-  "log": string, // a brief preamble to the user explaining what you’re about to do
+  "log": string, // a preamble message to the user explaining what you’re about to do
   ${commonOutputFields}
   "action": 
     {
@@ -133,7 +141,7 @@ Return in JSON format:
 For example, if the instruction is to login and the form has already been filled, this is a valid return value:
 
 {
-  "log": "Click the login button",
+  "log": "${logSample}",
   "more_actions_needed_by_instruction": false,
   "action": {
     "type": "Tap",

package/dist/lib/ai-model/prompt/llm-planning.js.map

@@ -1 +1 @@
-{"version":3,"file":"ai-model/prompt/llm-planning.js","sources":["webpack/runtime/define_property_getters","webpack/runtime/has_own_property","webpack/runtime/make_namespace_object","../../../../src/ai-model/prompt/llm-planning.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 { DeviceAction } from '@/types';\nimport type { TVlModeTypes } from '@midscene/shared/env';\nimport {\n  getZodDescription,\n  getZodTypeName,\n} from '@midscene/shared/zod-schema-utils';\nimport type { ResponseFormatJSONSchema } from 'openai/resources/index';\nimport type { z } from 'zod';\nimport { bboxDescription } from './common';\n\n// Note: put the log field first to trigger the CoT\n\nconst commonOutputFields = `\"error\"?: string, // Error messages about unexpected situations, if any. Only think it is an error when the situation is not foreseeable according to the instruction. Use the same language as the user's instruction.\n  \"more_actions_needed_by_instruction\": boolean, // Consider if there is still more action(s) to do after the action in \"Log\" is done, according to the instruction. If so, set this field to true. Otherwise, set it to false.`;\n\nconst vlLocateParam = (vlMode: TVlModeTypes | undefined) => {\n  if (vlMode) {\n    return `{bbox: [number, number, number, number], prompt: string } // ${bboxDescription(vlMode)}`;\n  }\n  return '{ prompt: string /* description of the target element */ }';\n};\n\nexport const descriptionForAction = (\n  action: DeviceAction<any>,\n  locatorSchemaTypeDescription: string,\n) => {\n  const tab = '  ';\n  const fields: string[] = [];\n\n  // Add the action type field\n  fields.push(`- type: \"${action.name}\"`);\n\n  // Handle paramSchema if it exists\n  if (action.paramSchema) {\n    const paramLines: string[] = [];\n\n    // Check if paramSchema is a ZodObject with shape\n    const schema = action.paramSchema as {\n      _def?: { typeName?: string };\n      shape?: Record<string, unknown>;\n    };\n    const isZodObject = schema._def?.typeName === 'ZodObject';\n\n    if (isZodObject && schema.shape) {\n      // Original logic for ZodObject schemas\n      const shape = schema.shape;\n\n      for (const [key, field] of Object.entries(shape)) {\n        if (field && typeof field === 'object') {\n          // Check if field is optional\n          const isOptional =\n            typeof (field as { isOptional?: () => boolean }).isOptional ===\n              'function' &&\n            (field as { isOptional: () => boolean }).isOptional();\n          const keyWithOptional = isOptional ? `${key}?` : key;\n\n          // Get the type name using extracted helper\n          const typeName = getZodTypeName(field, locatorSchemaTypeDescription);\n\n          // Get description using extracted helper\n          const description = getZodDescription(field as z.ZodTypeAny);\n\n          // Build param line for this field\n          let paramLine = `${keyWithOptional}: ${typeName}`;\n          if (description) {\n            paramLine += ` // ${description}`;\n          }\n\n          paramLines.push(paramLine);\n        }\n      }\n\n      // Add the param section to fields if there are paramLines\n      if (paramLines.length > 0) {\n        fields.push('- param:');\n        paramLines.forEach((line) => {\n          fields.push(`  - ${line}`);\n        });\n      }\n    } else {\n      // Handle non-object schemas (string, number, etc.)\n      const typeName = getZodTypeName(schema);\n      const description = getZodDescription(schema as z.ZodTypeAny);\n\n      // For simple types, indicate that param should be the direct value, not an object\n      let paramDescription = `- param: ${typeName}`;\n      if (description) {\n        paramDescription += ` // ${description}`;\n      }\n      paramDescription += ' (pass the value directly, not as an object)';\n\n      fields.push(paramDescription);\n    }\n  }\n\n  return `- ${action.name}, ${action.description || 'No description provided'}\n${tab}${fields.join(`\\n${tab}`)}\n`.trim();\n};\n\nexport async function systemPromptToTaskPlanning({\n  actionSpace,\n  vlMode,\n  includeBbox,\n}: {\n  actionSpace: DeviceAction<any>[];\n  vlMode: TVlModeTypes | undefined;\n  includeBbox: boolean;\n}) {\n  // Validate parameters: if includeBbox is true, vlMode must be defined\n  if (includeBbox && !vlMode) {\n    throw new Error(\n      'vlMode cannot be undefined when includeBbox is true. A valid vlMode is required for bbox-based location.',\n    );\n  }\n\n  const actionDescriptionList = actionSpace.map((action) => {\n    return descriptionForAction(\n      action,\n      vlLocateParam(includeBbox ? vlMode : undefined),\n    );\n  });\n  const actionList = actionDescriptionList.join('\\n');\n\n  const logFieldInstruction = `\n## About the \\`log\\` field (preamble message)\n\nThe \\`log\\` field is a brief preamble message to the user explaining what you’re about to do. It should follow these principles and examples:\n\n- **Use the same language as the user's instruction**\n- **Keep it concise**: be no more than 1-2 sentences, focused on immediate, tangible next steps. (8–12 words or Chinese characters for quick updates).\n- **Build on prior context**: if this is not the first action to be done, use the preamble message to connect the dots with what’s been done so far and create a sense of momentum and clarity for the user to understand your next actions.\n- **Keep your tone light, friendly and curious**: add small touches of personality in preambles feel collaborative and engaging.\n\n**Examples:**\n- \"Click the login button\"\n- \"Scroll to find the 'Yes' button in popup\"\n- \"Previous actions failed to find the 'Yes' button, i will try again\"\n- \"Go back to find the login button\"\n`;\n\n  return `\nTarget: User will give you an instruction, some screenshots and previous logs indicating what have been done. Your task is to plan the next one action according to current situation to accomplish the instruction.\n\nPlease tell what the next one action is (or null if no action should be done) to do the tasks the instruction requires. \n\n## Rules\n\n- Don't give extra actions or plans beyond the instruction. For example, don't try to submit the form if the instruction is only to fill something.\n- Give just the next ONE action you should do\n- Consider the current screenshot and give the action that is most likely to accomplish the instruction. For example, if the next step is to click a button but it's not visible in the screenshot, you should try to find it first instead of give a click action.\n- Make sure the previous actions are completed successfully before performing the next step\n- If there are some error messages reported by the previous actions, don't give up, try parse a new action to recover. If the error persists for more than 5 times, you should think this is an error and set the \"error\" field to the error message.\n- If there is nothing to do but waiting, set the \"sleep\" field to the positive waiting time in milliseconds and null for the \"action\" field.\n- Assertions are also important steps. When getting the assertion instruction, a solid conclusion is required. You should explicitly state your conclusion by calling the \"Print_Assert_Result\" action.\n\n## Supporting actions\n${actionList}\n\n${logFieldInstruction}\n\n## Return format\n\nReturn in JSON format:\n{\n  \"log\": string, // a brief preamble to the user explaining what you’re about to do\n  ${commonOutputFields}\n  \"action\": \n    {\n      \"type\": string, // the type of the action\n      \"param\"?: { // The parameter of the action, if any\n         // k-v style parameter fields\n      }, \n    } | null,\n  ,\n  \"sleep\"?: number, // The sleep time after the action, in milliseconds.\n}\n\nFor example, if the instruction is to login and the form has already been filled, this is a valid return value:\n\n{\n  \"log\": \"Click the login button\",\n  \"more_actions_needed_by_instruction\": false,\n  \"action\": {\n    \"type\": \"Tap\",\n    \"param\": {\n      \"locate\": { \n        \"prompt\": \"The login button\"${vlMode && includeBbox ? `, \"bbox\": [100, 200, 300, 400]` : ''}\n      }\n    }\n  }\n}\n`;\n}\n"],"names":["__webpack_require__","definition","key","Object","obj","prop","Symbol","commonOutputFields","vlLocateParam","vlMode","bboxDescription","descriptionForAction","action","locatorSchemaTypeDescription","tab","fields","paramLines","schema","isZodObject","shape","field","isOptional","keyWithOptional","typeName","getZodTypeName","description","getZodDescription","paramLine","line","paramDescription","systemPromptToTaskPlanning","actionSpace","includeBbox","Error","actionDescriptionList","undefined","actionList","logFieldInstruction"],"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;;;;;;;;;;ACMA,MAAMI,qBAAqB,CAAC;+NACmM,CAAC;AAEhO,MAAMC,gBAAgB,CAACC;IACrB,IAAIA,QACF,OAAO,CAAC,6DAA6D,EAAEC,AAAAA,IAAAA,mCAAAA,eAAAA,AAAAA,EAAgBD,SAAS;IAElG,OAAO;AACT;AAEO,MAAME,uBAAuB,CAClCC,QACAC;IAEA,MAAMC,MAAM;IACZ,MAAMC,SAAmB,EAAE;IAG3BA,OAAO,IAAI,CAAC,CAAC,SAAS,EAAEH,OAAO,IAAI,CAAC,CAAC,CAAC;IAGtC,IAAIA,OAAO,WAAW,EAAE;QACtB,MAAMI,aAAuB,EAAE;QAG/B,MAAMC,SAASL,OAAO,WAAW;QAIjC,MAAMM,cAAcD,OAAO,IAAI,EAAE,aAAa;QAE9C,IAAIC,eAAeD,OAAO,KAAK,EAAE;YAE/B,MAAME,QAAQF,OAAO,KAAK;YAE1B,KAAK,MAAM,CAACf,KAAKkB,MAAM,IAAIjB,OAAO,OAAO,CAACgB,OACxC,IAAIC,SAAS,AAAiB,YAAjB,OAAOA,OAAoB;gBAEtC,MAAMC,aACJ,AACE,cADF,OAAQD,MAAyC,UAAU,IAE1DA,MAAwC,UAAU;gBACrD,MAAME,kBAAkBD,aAAa,GAAGnB,IAAI,CAAC,CAAC,GAAGA;gBAGjD,MAAMqB,WAAWC,AAAAA,IAAAA,iCAAAA,cAAAA,AAAAA,EAAeJ,OAAOP;gBAGvC,MAAMY,cAAcC,AAAAA,IAAAA,iCAAAA,iBAAAA,AAAAA,EAAkBN;gBAGtC,IAAIO,YAAY,GAAGL,gBAAgB,EAAE,EAAEC,UAAU;gBACjD,IAAIE,aACFE,aAAa,CAAC,IAAI,EAAEF,aAAa;gBAGnCT,WAAW,IAAI,CAACW;YAClB;YAIF,IAAIX,WAAW,MAAM,GAAG,GAAG;gBACzBD,OAAO,IAAI,CAAC;gBACZC,WAAW,OAAO,CAAC,CAACY;oBAClBb,OAAO,IAAI,CAAC,CAAC,IAAI,EAAEa,MAAM;gBAC3B;YACF;QACF,OAAO;YAEL,MAAML,WAAWC,AAAAA,IAAAA,iCAAAA,cAAAA,AAAAA,EAAeP;YAChC,MAAMQ,cAAcC,AAAAA,IAAAA,iCAAAA,iBAAAA,AAAAA,EAAkBT;YAGtC,IAAIY,mBAAmB,CAAC,SAAS,EAAEN,UAAU;YAC7C,IAAIE,aACFI,oBAAoB,CAAC,IAAI,EAAEJ,aAAa;YAE1CI,oBAAoB;YAEpBd,OAAO,IAAI,CAACc;QACd;IACF;IAEA,OAAO,CAAC,EAAE,EAAEjB,OAAO,IAAI,CAAC,EAAE,EAAEA,OAAO,WAAW,IAAI,0BAA0B;AAC9E,EAAEE,MAAMC,OAAO,IAAI,CAAC,CAAC,EAAE,EAAED,KAAK,EAAE;AAChC,CAAC,CAAC,IAAI;AACN;AAEO,eAAegB,2BAA2B,EAC/CC,WAAW,EACXtB,MAAM,EACNuB,WAAW,EAKZ;IAEC,IAAIA,eAAe,CAACvB,QAClB,MAAM,IAAIwB,MACR;IAIJ,MAAMC,wBAAwBH,YAAY,GAAG,CAAC,CAACnB,SACtCD,qBACLC,QACAJ,cAAcwB,cAAcvB,SAAS0B;IAGzC,MAAMC,aAAaF,sBAAsB,IAAI,CAAC;IAE9C,MAAMG,sBAAsB,CAAC;;;;;;;;;;;;;;;AAe/B,CAAC;IAEC,OAAO,CAAC;;;;;;;;;;;;;;;;AAgBV,EAAED,WAAW;;AAEb,EAAEC,oBAAoB;;;;;;;EAOpB,EAAE9B,mBAAmB;;;;;;;;;;;;;;;;;;;;;oCAqBa,EAAEE,UAAUuB,cAAc,mCAAmC,GAAG;;;;;AAKpG,CAAC;AACD"}
+{"version":3,"file":"ai-model/prompt/llm-planning.js","sources":["webpack/runtime/define_property_getters","webpack/runtime/has_own_property","webpack/runtime/make_namespace_object","../../../../src/ai-model/prompt/llm-planning.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 { DeviceAction } from '@/types';\nimport type { TVlModeTypes } from '@midscene/shared/env';\nimport {\n  getZodDescription,\n  getZodTypeName,\n} from '@midscene/shared/zod-schema-utils';\nimport type { ResponseFormatJSONSchema } from 'openai/resources/index';\nimport type { z } from 'zod';\nimport { bboxDescription } from './common';\n\n// Note: put the log field first to trigger the CoT\n\nconst commonOutputFields = `\"error\"?: string, // Error messages about unexpected situations, if any. Only think it is an error when the situation is not foreseeable according to the instruction. Use the same language as the user's instruction.\n  \"more_actions_needed_by_instruction\": boolean, // Consider if there is still more action(s) to do after the action in \"Log\" is done, according to the instruction. If so, set this field to true. Otherwise, set it to false.`;\n\nconst vlLocateParam = (vlMode: TVlModeTypes | undefined) => {\n  if (vlMode) {\n    return `{bbox: [number, number, number, number], prompt: string } // ${bboxDescription(vlMode)}`;\n  }\n  return '{ prompt: string /* description of the target element */ }';\n};\n\nexport const descriptionForAction = (\n  action: DeviceAction<any>,\n  locatorSchemaTypeDescription: string,\n) => {\n  const tab = '  ';\n  const fields: string[] = [];\n\n  // Add the action type field\n  fields.push(`- type: \"${action.name}\"`);\n\n  // Handle paramSchema if it exists\n  if (action.paramSchema) {\n    const paramLines: string[] = [];\n\n    // Check if paramSchema is a ZodObject with shape\n    const schema = action.paramSchema as {\n      _def?: { typeName?: string };\n      shape?: Record<string, unknown>;\n    };\n    const isZodObject = schema._def?.typeName === 'ZodObject';\n\n    if (isZodObject && schema.shape) {\n      // Original logic for ZodObject schemas\n      const shape = schema.shape;\n\n      for (const [key, field] of Object.entries(shape)) {\n        if (field && typeof field === 'object') {\n          // Check if field is optional\n          const isOptional =\n            typeof (field as { isOptional?: () => boolean }).isOptional ===\n              'function' &&\n            (field as { isOptional: () => boolean }).isOptional();\n          const keyWithOptional = isOptional ? `${key}?` : key;\n\n          // Get the type name using extracted helper\n          const typeName = getZodTypeName(field, locatorSchemaTypeDescription);\n\n          // Get description using extracted helper\n          const description = getZodDescription(field as z.ZodTypeAny);\n\n          // Build param line for this field\n          let paramLine = `${keyWithOptional}: ${typeName}`;\n          if (description) {\n            paramLine += ` // ${description}`;\n          }\n\n          paramLines.push(paramLine);\n        }\n      }\n\n      // Add the param section to fields if there are paramLines\n      if (paramLines.length > 0) {\n        fields.push('- param:');\n        paramLines.forEach((line) => {\n          fields.push(`  - ${line}`);\n        });\n      }\n    } else {\n      // Handle non-object schemas (string, number, etc.)\n      const typeName = getZodTypeName(schema);\n      const description = getZodDescription(schema as z.ZodTypeAny);\n\n      // For simple types, indicate that param should be the direct value, not an object\n      let paramDescription = `- param: ${typeName}`;\n      if (description) {\n        paramDescription += ` // ${description}`;\n      }\n      paramDescription += ' (pass the value directly, not as an object)';\n\n      fields.push(paramDescription);\n    }\n  }\n\n  return `- ${action.name}, ${action.description || 'No description provided'}\n${tab}${fields.join(`\\n${tab}`)}\n`.trim();\n};\n\nexport async function systemPromptToTaskPlanning({\n  actionSpace,\n  vlMode,\n  includeBbox,\n}: {\n  actionSpace: DeviceAction<any>[];\n  vlMode: TVlModeTypes | undefined;\n  includeBbox: boolean;\n}) {\n  // Validate parameters: if includeBbox is true, vlMode must be defined\n  if (includeBbox && !vlMode) {\n    throw new Error(\n      'vlMode cannot be undefined when includeBbox is true. A valid vlMode is required for bbox-based location.',\n    );\n  }\n\n  const actionDescriptionList = actionSpace.map((action) => {\n    return descriptionForAction(\n      action,\n      vlLocateParam(includeBbox ? vlMode : undefined),\n    );\n  });\n  const actionList = actionDescriptionList.join('\\n');\n\n  const logFieldInstruction =\n    vlMode === 'qwen2.5-vl' || vlMode === 'qwen3-vl'\n      ? `\n## About the \\`log\\` field\n\n* Log your thoughts and what the next one action (ONLY ONE!) you can do according to the screenshot and the instruction. \n* The log should contain the following information: \"The user wants to do ... . According to the instruction and the previous logs, next step is to .... Now i am going to compose an action '{ action-type }' to do ....\". \n* If no action should be done, log the reason. \n* Use the same language as the user's instruction.\n`\n      : `\n## About the \\`log\\` field (preamble message)\n\nThe \\`log\\` field is a brief preamble message to the user explaining what you’re about to do. It should follow these principles and examples:\n\n- **Use the same language as the user's instruction**\n- **Keep it concise**: be no more than 1-2 sentences, focused on immediate, tangible next steps. (8–12 words or Chinese characters for quick updates).\n- **Build on prior context**: if this is not the first action to be done, use the preamble message to connect the dots with what’s been done so far and create a sense of momentum and clarity for the user to understand your next actions.\n- **Keep your tone light, friendly and curious**: add small touches of personality in preambles feel collaborative and engaging.\n\n**Examples:**\n- \"Click the login button\"\n- \"Scroll to find the 'Yes' button in popup\"\n- \"Previous actions failed to find the 'Yes' button, i will try again\"\n- \"Go back to find the login button\"\n`;\n\n  const logSample =\n    vlMode === 'qwen2.5-vl' || vlMode === 'qwen3-vl'\n      ? 'The user wants to fill the form and login. According to the instruction and the previous logs, next step is to click the login button. Now i am going to compose an action \"Tap\" to click the login button.'\n      : 'Click the login button';\n\n  return `\nTarget: User will give you an instruction, some screenshots and previous logs indicating what have been done. Your task is to plan the next one action according to current situation to accomplish the instruction.\n\nPlease tell what the next one action is (or null if no action should be done) to do the tasks the instruction requires. \n\n## Rules\n\n- Don't give extra actions or plans beyond the instruction. For example, don't try to submit the form if the instruction is only to fill something.\n- Give just the next ONE action you should do\n- Consider the current screenshot and give the action that is most likely to accomplish the instruction. For example, if the next step is to click a button but it's not visible in the screenshot, you should try to find it first instead of give a click action.\n- Make sure the previous actions are completed successfully before performing the next step\n- If there are some error messages reported by the previous actions, don't give up, try parse a new action to recover. If the error persists for more than 5 times, you should think this is an error and set the \"error\" field to the error message.\n- If there is nothing to do but waiting, set the \"sleep\" field to the positive waiting time in milliseconds and null for the \"action\" field.\n- Assertions are also important steps. When getting the assertion instruction, a solid conclusion is required. You should explicitly state your conclusion by calling the \"Print_Assert_Result\" action.\n\n## Supporting actions\n${actionList}\n\n${logFieldInstruction}\n\n## Return format\n\nReturn in JSON format:\n{\n  \"log\": string, // a preamble message to the user explaining what you’re about to do\n  ${commonOutputFields}\n  \"action\": \n    {\n      \"type\": string, // the type of the action\n      \"param\"?: { // The parameter of the action, if any\n         // k-v style parameter fields\n      }, \n    } | null,\n  ,\n  \"sleep\"?: number, // The sleep time after the action, in milliseconds.\n}\n\nFor example, if the instruction is to login and the form has already been filled, this is a valid return value:\n\n{\n  \"log\": \"${logSample}\",\n  \"more_actions_needed_by_instruction\": false,\n  \"action\": {\n    \"type\": \"Tap\",\n    \"param\": {\n      \"locate\": { \n        \"prompt\": \"The login button\"${vlMode && includeBbox ? `, \"bbox\": [100, 200, 300, 400]` : ''}\n      }\n    }\n  }\n}\n`;\n}\n"],"names":["__webpack_require__","definition","key","Object","obj","prop","Symbol","commonOutputFields","vlLocateParam","vlMode","bboxDescription","descriptionForAction","action","locatorSchemaTypeDescription","tab","fields","paramLines","schema","isZodObject","shape","field","isOptional","keyWithOptional","typeName","getZodTypeName","description","getZodDescription","paramLine","line","paramDescription","systemPromptToTaskPlanning","actionSpace","includeBbox","Error","actionDescriptionList","undefined","actionList","logFieldInstruction","logSample"],"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;;;;;;;;;;ACMA,MAAMI,qBAAqB,CAAC;+NACmM,CAAC;AAEhO,MAAMC,gBAAgB,CAACC;IACrB,IAAIA,QACF,OAAO,CAAC,6DAA6D,EAAEC,AAAAA,IAAAA,mCAAAA,eAAAA,AAAAA,EAAgBD,SAAS;IAElG,OAAO;AACT;AAEO,MAAME,uBAAuB,CAClCC,QACAC;IAEA,MAAMC,MAAM;IACZ,MAAMC,SAAmB,EAAE;IAG3BA,OAAO,IAAI,CAAC,CAAC,SAAS,EAAEH,OAAO,IAAI,CAAC,CAAC,CAAC;IAGtC,IAAIA,OAAO,WAAW,EAAE;QACtB,MAAMI,aAAuB,EAAE;QAG/B,MAAMC,SAASL,OAAO,WAAW;QAIjC,MAAMM,cAAcD,OAAO,IAAI,EAAE,aAAa;QAE9C,IAAIC,eAAeD,OAAO,KAAK,EAAE;YAE/B,MAAME,QAAQF,OAAO,KAAK;YAE1B,KAAK,MAAM,CAACf,KAAKkB,MAAM,IAAIjB,OAAO,OAAO,CAACgB,OACxC,IAAIC,SAAS,AAAiB,YAAjB,OAAOA,OAAoB;gBAEtC,MAAMC,aACJ,AACE,cADF,OAAQD,MAAyC,UAAU,IAE1DA,MAAwC,UAAU;gBACrD,MAAME,kBAAkBD,aAAa,GAAGnB,IAAI,CAAC,CAAC,GAAGA;gBAGjD,MAAMqB,WAAWC,AAAAA,IAAAA,iCAAAA,cAAAA,AAAAA,EAAeJ,OAAOP;gBAGvC,MAAMY,cAAcC,AAAAA,IAAAA,iCAAAA,iBAAAA,AAAAA,EAAkBN;gBAGtC,IAAIO,YAAY,GAAGL,gBAAgB,EAAE,EAAEC,UAAU;gBACjD,IAAIE,aACFE,aAAa,CAAC,IAAI,EAAEF,aAAa;gBAGnCT,WAAW,IAAI,CAACW;YAClB;YAIF,IAAIX,WAAW,MAAM,GAAG,GAAG;gBACzBD,OAAO,IAAI,CAAC;gBACZC,WAAW,OAAO,CAAC,CAACY;oBAClBb,OAAO,IAAI,CAAC,CAAC,IAAI,EAAEa,MAAM;gBAC3B;YACF;QACF,OAAO;YAEL,MAAML,WAAWC,AAAAA,IAAAA,iCAAAA,cAAAA,AAAAA,EAAeP;YAChC,MAAMQ,cAAcC,AAAAA,IAAAA,iCAAAA,iBAAAA,AAAAA,EAAkBT;YAGtC,IAAIY,mBAAmB,CAAC,SAAS,EAAEN,UAAU;YAC7C,IAAIE,aACFI,oBAAoB,CAAC,IAAI,EAAEJ,aAAa;YAE1CI,oBAAoB;YAEpBd,OAAO,IAAI,CAACc;QACd;IACF;IAEA,OAAO,CAAC,EAAE,EAAEjB,OAAO,IAAI,CAAC,EAAE,EAAEA,OAAO,WAAW,IAAI,0BAA0B;AAC9E,EAAEE,MAAMC,OAAO,IAAI,CAAC,CAAC,EAAE,EAAED,KAAK,EAAE;AAChC,CAAC,CAAC,IAAI;AACN;AAEO,eAAegB,2BAA2B,EAC/CC,WAAW,EACXtB,MAAM,EACNuB,WAAW,EAKZ;IAEC,IAAIA,eAAe,CAACvB,QAClB,MAAM,IAAIwB,MACR;IAIJ,MAAMC,wBAAwBH,YAAY,GAAG,CAAC,CAACnB,SACtCD,qBACLC,QACAJ,cAAcwB,cAAcvB,SAAS0B;IAGzC,MAAMC,aAAaF,sBAAsB,IAAI,CAAC;IAE9C,MAAMG,sBACJ5B,AAAW,iBAAXA,UAA2BA,AAAW,eAAXA,SACvB,CAAC;;;;;;;AAOT,CAAC,GACO,CAAC;;;;;;;;;;;;;;;AAeT,CAAC;IAEC,MAAM6B,YACJ7B,AAAW,iBAAXA,UAA2BA,AAAW,eAAXA,SACvB,gNACA;IAEN,OAAO,CAAC;;;;;;;;;;;;;;;;AAgBV,EAAE2B,WAAW;;AAEb,EAAEC,oBAAoB;;;;;;;EAOpB,EAAE9B,mBAAmB;;;;;;;;;;;;;;;UAeb,EAAE+B,UAAU;;;;;;oCAMc,EAAE7B,UAAUuB,cAAc,mCAAmC,GAAG;;;;;AAKpG,CAAC;AACD"}

package/dist/lib/device/index.js

@@ -129,12 +129,6 @@ const defineActionKeyboardPress = (call)=>defineAction({
         call
     });
 const actionScrollParamSchema = external_zod_namespaceObject.z.object({
-    direction: external_zod_namespaceObject.z["enum"]([
-        'down',
-        'up',
-        'right',
-        'left'
-    ]).default('down').describe('The direction to scroll'),
     scrollType: external_zod_namespaceObject.z["enum"]([
         'singleAction',
         'scrollToBottom',
@@ -142,6 +136,12 @@ const actionScrollParamSchema = external_zod_namespaceObject.z.object({
         'scrollToRight',
         'scrollToLeft'
     ]).default('singleAction').describe('The scroll behavior: "singleAction" for a single scroll action, "scrollToBottom" for scrolling to the bottom, "scrollToTop" for scrolling to the top, "scrollToRight" for scrolling to the right, "scrollToLeft" for scrolling to the left'),
+    direction: external_zod_namespaceObject.z["enum"]([
+        'down',
+        'up',
+        'right',
+        'left'
+    ]).default('down').describe('The direction to scroll. Only effective when scrollType is "singleAction".'),
     distance: external_zod_namespaceObject.z.number().nullable().optional().describe('The distance in pixels to scroll'),
     locate: (0, external_common_js_namespaceObject.getMidsceneLocationSchema)().optional().describe('The target element to be scrolled on, like "the table" or "the list" or "the content area" or "the scrollable area"')
 });

package/dist/lib/device/index.js.map

@@ -1 +1 @@
-{"version":3,"file":"device/index.js","sources":["webpack/runtime/define_property_getters","webpack/runtime/has_own_property","webpack/runtime/make_namespace_object","../../../src/device/index.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 { getMidsceneLocationSchema } from '@/common';\nimport type {\n  ActionScrollParam,\n  DeviceAction,\n  LocateResultElement,\n} from '@/types';\nimport type { IModelConfig } from '@midscene/shared/env';\nimport type { ElementNode } from '@midscene/shared/extractor';\nimport { getDebug } from '@midscene/shared/logger';\nimport { _keyDefinitions } from '@midscene/shared/us-keyboard-layout';\nimport { z } from 'zod';\nimport type { ElementCacheFeature, Rect, Size, UIContext } from '../types';\n\nexport abstract class AbstractInterface {\n  abstract interfaceType: string;\n\n  abstract screenshotBase64(): Promise<string>;\n  abstract size(): Promise<Size>;\n  abstract actionSpace(): DeviceAction[];\n\n  abstract cacheFeatureForRect?(\n    rect: Rect,\n    options?: {\n      targetDescription?: string;\n      modelConfig?: IModelConfig;\n    },\n  ): Promise<ElementCacheFeature>;\n  abstract rectMatchesCacheFeature?(\n    feature: ElementCacheFeature,\n  ): Promise<Rect>;\n\n  abstract destroy?(): Promise<void>;\n\n  abstract describe?(): string;\n  abstract beforeInvokeAction?(actionName: string, param: any): Promise<void>;\n  abstract afterInvokeAction?(actionName: string, param: any): Promise<void>;\n\n  // @deprecated do NOT extend this method\n  abstract getElementsNodeTree?: () => Promise<ElementNode>;\n\n  // @deprecated do NOT extend this method\n  abstract url?: () => string | Promise<string>;\n\n  // @deprecated do NOT extend this method\n  abstract evaluateJavaScript?<T = any>(script: string): Promise<T>;\n\n  // @deprecated do NOT extend this method\n  abstract getContext?(): Promise<UIContext>;\n}\n\n// Generic function to define actions with proper type inference\n// TRuntime allows specifying a different type for the runtime parameter (after location resolution)\n// TReturn allows specifying the return type of the action\nexport const defineAction = <\n  TSchema extends z.ZodType | undefined = undefined,\n  TRuntime = TSchema extends z.ZodType ? z.infer<TSchema> : undefined,\n  TReturn = any,\n>(\n  config: {\n    name: string;\n    description: string;\n    interfaceAlias?: string;\n    paramSchema?: TSchema;\n    call: (param: TRuntime) => Promise<TReturn> | TReturn;\n  } & Partial<\n    Omit<\n      DeviceAction<TRuntime, TReturn>,\n      'name' | 'description' | 'interfaceAlias' | 'paramSchema' | 'call'\n    >\n  >,\n): DeviceAction<TRuntime, TReturn> => {\n  return config as any; // Type assertion needed because schema validation type differs from runtime type\n};\n\n// Tap\nexport const actionTapParamSchema = z.object({\n  locate: getMidsceneLocationSchema().describe('The element to be tapped'),\n});\n// Override the inferred type to use LocateResultElement for the runtime locate field\nexport type ActionTapParam = {\n  locate: LocateResultElement;\n};\n\nexport const defineActionTap = (\n  call: (param: ActionTapParam) => Promise<void>,\n): DeviceAction<ActionTapParam> => {\n  return defineAction<typeof actionTapParamSchema, ActionTapParam>({\n    name: 'Tap',\n    description: 'Tap the element',\n    interfaceAlias: 'aiTap',\n    paramSchema: actionTapParamSchema,\n    call,\n  });\n};\n\n// RightClick\nexport const actionRightClickParamSchema = z.object({\n  locate: getMidsceneLocationSchema().describe(\n    'The element to be right clicked',\n  ),\n});\nexport type ActionRightClickParam = {\n  locate: LocateResultElement;\n};\n\nexport const defineActionRightClick = (\n  call: (param: ActionRightClickParam) => Promise<void>,\n): DeviceAction<ActionRightClickParam> => {\n  return defineAction<\n    typeof actionRightClickParamSchema,\n    ActionRightClickParam\n  >({\n    name: 'RightClick',\n    description: 'Right click the element',\n    interfaceAlias: 'aiRightClick',\n    paramSchema: actionRightClickParamSchema,\n    call,\n  });\n};\n\n// DoubleClick\nexport const actionDoubleClickParamSchema = z.object({\n  locate: getMidsceneLocationSchema().describe(\n    'The element to be double clicked',\n  ),\n});\nexport type ActionDoubleClickParam = {\n  locate: LocateResultElement;\n};\n\nexport const defineActionDoubleClick = (\n  call: (param: ActionDoubleClickParam) => Promise<void>,\n): DeviceAction<ActionDoubleClickParam> => {\n  return defineAction<\n    typeof actionDoubleClickParamSchema,\n    ActionDoubleClickParam\n  >({\n    name: 'DoubleClick',\n    description: 'Double click the element',\n    interfaceAlias: 'aiDoubleClick',\n    paramSchema: actionDoubleClickParamSchema,\n    call,\n  });\n};\n\n// Hover\nexport const actionHoverParamSchema = z.object({\n  locate: getMidsceneLocationSchema().describe('The element to be hovered'),\n});\nexport type ActionHoverParam = {\n  locate: LocateResultElement;\n};\n\nexport const defineActionHover = (\n  call: (param: ActionHoverParam) => Promise<void>,\n): DeviceAction<ActionHoverParam> => {\n  return defineAction<typeof actionHoverParamSchema, ActionHoverParam>({\n    name: 'Hover',\n    description: 'Move the mouse to the element',\n    interfaceAlias: 'aiHover',\n    paramSchema: actionHoverParamSchema,\n    call,\n  });\n};\n\n// Input\nconst inputLocateDescription =\n  'the position of the placeholder or text content in the target input field. If there is no content, locate the center of the input field.';\nexport const actionInputParamSchema = z.object({\n  value: z\n    .union([z.string(), z.number()])\n    .transform((val) => String(val))\n    .describe(\n      'The text to input. Provide the final content for replace/append modes, or an empty string when using clear mode to remove existing text.',\n    ),\n  locate: getMidsceneLocationSchema()\n    .describe(inputLocateDescription)\n    .optional(),\n  mode: z\n    .enum(['replace', 'clear', 'append'])\n    .default('replace')\n    .optional()\n    .describe(\n      'Input mode: \"replace\" (default) - clear the field and input the value; \"append\" - append the value to existing content; \"clear\" - clear the field without inputting new text.',\n    ),\n});\nexport type ActionInputParam = {\n  value: string;\n  locate?: LocateResultElement;\n  mode?: 'replace' | 'clear' | 'append';\n};\n\nexport const defineActionInput = (\n  call: (param: ActionInputParam) => Promise<void>,\n): DeviceAction<ActionInputParam> => {\n  return defineAction<typeof actionInputParamSchema, ActionInputParam>({\n    name: 'Input',\n    description: 'Input the value into the element',\n    interfaceAlias: 'aiInput',\n    paramSchema: actionInputParamSchema,\n    call,\n  });\n};\n\n// KeyboardPress\nexport const actionKeyboardPressParamSchema = z.object({\n  locate: getMidsceneLocationSchema()\n    .describe('The element to be clicked before pressing the key')\n    .optional(),\n  keyName: z\n    .string()\n    .describe(\n      \"The key to be pressed. Use '+' for key combinations, e.g., 'Control+A', 'Shift+Enter'\",\n    ),\n});\nexport type ActionKeyboardPressParam = {\n  locate?: LocateResultElement;\n  keyName: string;\n};\n\nexport const defineActionKeyboardPress = (\n  call: (param: ActionKeyboardPressParam) => Promise<void>,\n): DeviceAction<ActionKeyboardPressParam> => {\n  return defineAction<\n    typeof actionKeyboardPressParamSchema,\n    ActionKeyboardPressParam\n  >({\n    name: 'KeyboardPress',\n    description:\n      'Press a key or key combination, like \"Enter\", \"Tab\", \"Escape\", or \"Control+A\", \"Shift+Enter\". Do not use this to type text.',\n    interfaceAlias: 'aiKeyboardPress',\n    paramSchema: actionKeyboardPressParamSchema,\n    call,\n  });\n};\n\n// Scroll\nexport const actionScrollParamSchema = z.object({\n  direction: z\n    .enum(['down', 'up', 'right', 'left'])\n    .default('down')\n    .describe('The direction to scroll'),\n  scrollType: z\n    .enum([\n      'singleAction',\n      'scrollToBottom',\n      'scrollToTop',\n      'scrollToRight',\n      'scrollToLeft',\n    ])\n    .default('singleAction')\n    .describe(\n      'The scroll behavior: \"singleAction\" for a single scroll action, \"scrollToBottom\" for scrolling to the bottom, \"scrollToTop\" for scrolling to the top, \"scrollToRight\" for scrolling to the right, \"scrollToLeft\" for scrolling to the left',\n    ),\n  distance: z\n    .number()\n    .nullable()\n    .optional()\n    .describe('The distance in pixels to scroll'),\n  locate: getMidsceneLocationSchema()\n    .optional()\n    .describe(\n      'The target element to be scrolled on, like \"the table\" or \"the list\" or \"the content area\" or \"the scrollable area\"',\n    ),\n});\n\nexport const defineActionScroll = (\n  call: (param: ActionScrollParam) => Promise<void>,\n): DeviceAction<ActionScrollParam> => {\n  return defineAction<typeof actionScrollParamSchema, ActionScrollParam>({\n    name: 'Scroll',\n    description:\n      'Scroll the page or an element. The direction to scroll, the scroll type, and the distance to scroll. The distance is the number of pixels to scroll. If not specified, use `down` direction, `once` scroll type, and `null` distance.',\n    interfaceAlias: 'aiScroll',\n    paramSchema: actionScrollParamSchema,\n    call,\n  });\n};\n\n// DragAndDrop\nexport const actionDragAndDropParamSchema = z.object({\n  from: getMidsceneLocationSchema().describe('The position to be dragged'),\n  to: getMidsceneLocationSchema().describe('The position to be dropped'),\n});\nexport type ActionDragAndDropParam = {\n  from: LocateResultElement;\n  to: LocateResultElement;\n};\n\nexport const defineActionDragAndDrop = (\n  call: (param: ActionDragAndDropParam) => Promise<void>,\n): DeviceAction<ActionDragAndDropParam> => {\n  return defineAction<\n    typeof actionDragAndDropParamSchema,\n    ActionDragAndDropParam\n  >({\n    name: 'DragAndDrop',\n    description:\n      'Drag and drop (hold the mouse or finger down and move the mouse) ',\n    interfaceAlias: 'aiDragAndDrop',\n    paramSchema: actionDragAndDropParamSchema,\n    call,\n  });\n};\n\nexport const ActionLongPressParamSchema = z.object({\n  locate: getMidsceneLocationSchema().describe(\n    'The element to be long pressed',\n  ),\n  duration: z\n    .number()\n    .default(500)\n    .optional()\n    .describe('Long press duration in milliseconds'),\n});\n\nexport type ActionLongPressParam = {\n  locate: LocateResultElement;\n  duration?: number;\n};\nexport const defineActionLongPress = (\n  call: (param: ActionLongPressParam) => Promise<void>,\n): DeviceAction<ActionLongPressParam> => {\n  return defineAction<typeof ActionLongPressParamSchema, ActionLongPressParam>({\n    name: 'LongPress',\n    description: 'Long press the element',\n    paramSchema: ActionLongPressParamSchema,\n    call,\n  });\n};\n\nexport const ActionSwipeParamSchema = z.object({\n  start: getMidsceneLocationSchema()\n    .optional()\n    .describe(\n      'Starting point of the swipe gesture, if not specified, the center of the page will be used',\n    ),\n  direction: z\n    .enum(['up', 'down', 'left', 'right'])\n    .optional()\n    .describe(\n      'The direction to swipe (required when using distance). The direction means the direction of the finger swipe.',\n    ),\n  distance: z\n    .number()\n    .optional()\n    .describe('The distance in pixels to swipe (mutually exclusive with end)'),\n  end: getMidsceneLocationSchema()\n    .optional()\n    .describe(\n      'Ending point of the swipe gesture (mutually exclusive with distance)',\n    ),\n  duration: z\n    .number()\n    .default(300)\n    .describe('Duration of the swipe gesture in milliseconds'),\n  repeat: z\n    .number()\n    .optional()\n    .describe(\n      'The number of times to repeat the swipe gesture. 1 for default, 0 for infinite (e.g. endless swipe until the end of the page)',\n    ),\n});\n\nexport type ActionSwipeParam = {\n  start?: LocateResultElement;\n  direction?: 'up' | 'down' | 'left' | 'right';\n  distance?: number;\n  end?: LocateResultElement;\n  duration?: number;\n  repeat?: number;\n};\n\nexport const defineActionSwipe = (\n  call: (param: ActionSwipeParam) => Promise<void>,\n): DeviceAction<ActionSwipeParam> => {\n  return defineAction<typeof ActionSwipeParamSchema, ActionSwipeParam>({\n    name: 'Swipe',\n    description:\n      'Perform a swipe gesture. You must specify either \"end\" (target location) or \"distance\" + \"direction\" - they are mutually exclusive. Use \"end\" for precise location-based swipes, or \"distance\" + \"direction\" for relative movement.',\n    paramSchema: ActionSwipeParamSchema,\n    call,\n  });\n};\n\n// ClearInput\nexport const actionClearInputParamSchema = z.object({\n  locate: getMidsceneLocationSchema().describe('The input field to be cleared'),\n});\nexport type ActionClearInputParam = {\n  locate: LocateResultElement;\n};\n\nexport const defineActionClearInput = (\n  call: (param: ActionClearInputParam) => Promise<void>,\n): DeviceAction<ActionClearInputParam> => {\n  return defineAction<\n    typeof actionClearInputParamSchema,\n    ActionClearInputParam\n  >({\n    name: 'ClearInput',\n    description: inputLocateDescription,\n    interfaceAlias: 'aiClearInput',\n    paramSchema: actionClearInputParamSchema,\n    call,\n  });\n};\n\n// Assert\nexport const actionAssertParamSchema = z.object({\n  condition: z.string().describe('The condition of the assertion'),\n  thought: z\n    .string()\n    .describe(\n      'The thought of the assertion, like \"I can see there are A, B, C elements on the page, which means ... , so the assertion is true\"',\n    ),\n  result: z.boolean().describe('The result of the assertion, true or false'),\n});\nexport type ActionAssertParam = {\n  condition: string;\n  thought: string;\n  result: boolean;\n};\n\nexport const defineActionAssert = (): DeviceAction<ActionAssertParam> => {\n  return defineAction<typeof actionAssertParamSchema, ActionAssertParam>({\n    name: 'Print_Assert_Result',\n    description: 'Print the result of the assertion',\n    paramSchema: actionAssertParamSchema,\n    call: async (param) => {\n      if (typeof param?.result !== 'boolean') {\n        throw new Error(\n          `The result of the assertion must be a boolean, but got: ${typeof param?.result}. ${param.thought || '(no thought)'}`,\n        );\n      }\n\n      getDebug('device:common-action')(\n        `Assert: ${param.condition}, Thought: ${param.thought}, Result: ${param.result}`,\n      );\n\n      if (!param.result) {\n        throw new Error(\n          `Assertion failed: ${param.thought || '(no thought)'} (Assertion = ${param.condition})`,\n        );\n      }\n    },\n  });\n};\n\nexport type { DeviceAction } from '../types';\nexport type {\n  AndroidDeviceOpt,\n  AndroidDeviceInputOpt,\n  IOSDeviceOpt,\n  IOSDeviceInputOpt,\n} from './device-options';\n"],"names":["__webpack_require__","definition","key","Object","obj","prop","Symbol","AbstractInterface","defineAction","config","actionTapParamSchema","z","getMidsceneLocationSchema","defineActionTap","call","actionRightClickParamSchema","defineActionRightClick","actionDoubleClickParamSchema","defineActionDoubleClick","actionHoverParamSchema","defineActionHover","inputLocateDescription","actionInputParamSchema","val","String","defineActionInput","actionKeyboardPressParamSchema","defineActionKeyboardPress","actionScrollParamSchema","defineActionScroll","actionDragAndDropParamSchema","defineActionDragAndDrop","ActionLongPressParamSchema","defineActionLongPress","ActionSwipeParamSchema","defineActionSwipe","actionClearInputParamSchema","defineActionClearInput","actionAssertParamSchema","defineActionAssert","param","Error","getDebug"],"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;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;ACOO,MAAeI;AAmCtB;AAKO,MAAMC,eAAe,CAK1BC,SAaOA;AAIF,MAAMC,uBAAuBC,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IAC3C,QAAQC,AAAAA,IAAAA,mCAAAA,yBAAAA,AAAAA,IAA4B,QAAQ,CAAC;AAC/C;AAMO,MAAMC,kBAAkB,CAC7BC,OAEON,aAA0D;QAC/D,MAAM;QACN,aAAa;QACb,gBAAgB;QAChB,aAAaE;QACbI;IACF;AAIK,MAAMC,8BAA8BJ,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IAClD,QAAQC,AAAAA,IAAAA,mCAAAA,yBAAAA,AAAAA,IAA4B,QAAQ,CAC1C;AAEJ;AAKO,MAAMI,yBAAyB,CACpCF,OAEON,aAGL;QACA,MAAM;QACN,aAAa;QACb,gBAAgB;QAChB,aAAaO;QACbD;IACF;AAIK,MAAMG,+BAA+BN,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IACnD,QAAQC,AAAAA,IAAAA,mCAAAA,yBAAAA,AAAAA,IAA4B,QAAQ,CAC1C;AAEJ;AAKO,MAAMM,0BAA0B,CACrCJ,OAEON,aAGL;QACA,MAAM;QACN,aAAa;QACb,gBAAgB;QAChB,aAAaS;QACbH;IACF;AAIK,MAAMK,yBAAyBR,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IAC7C,QAAQC,AAAAA,IAAAA,mCAAAA,yBAAAA,AAAAA,IAA4B,QAAQ,CAAC;AAC/C;AAKO,MAAMQ,oBAAoB,CAC/BN,OAEON,aAA8D;QACnE,MAAM;QACN,aAAa;QACb,gBAAgB;QAChB,aAAaW;QACbL;IACF;AAIF,MAAMO,yBACJ;AACK,MAAMC,yBAAyBX,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IAC7C,OAAOA,6BAAAA,CAAAA,CAAAA,KACC,CAAC;QAACA,6BAAAA,CAAAA,CAAAA,MAAQ;QAAIA,6BAAAA,CAAAA,CAAAA,MAAQ;KAAG,EAC9B,SAAS,CAAC,CAACY,MAAQC,OAAOD,MAC1B,QAAQ,CACP;IAEJ,QAAQX,AAAAA,IAAAA,mCAAAA,yBAAAA,AAAAA,IACL,QAAQ,CAACS,wBACT,QAAQ;IACX,MAAMV,6BAAAA,CAAAA,CAAAA,OACC,CAAC;QAAC;QAAW;QAAS;KAAS,EACnC,OAAO,CAAC,WACR,QAAQ,GACR,QAAQ,CACP;AAEN;AAOO,MAAMc,oBAAoB,CAC/BX,OAEON,aAA8D;QACnE,MAAM;QACN,aAAa;QACb,gBAAgB;QAChB,aAAac;QACbR;IACF;AAIK,MAAMY,iCAAiCf,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IACrD,QAAQC,AAAAA,IAAAA,mCAAAA,yBAAAA,AAAAA,IACL,QAAQ,CAAC,qDACT,QAAQ;IACX,SAASD,6BAAAA,CAAAA,CAAAA,MACA,GACN,QAAQ,CACP;AAEN;AAMO,MAAMgB,4BAA4B,CACvCb,OAEON,aAGL;QACA,MAAM;QACN,aACE;QACF,gBAAgB;QAChB,aAAakB;QACbZ;IACF;AAIK,MAAMc,0BAA0BjB,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IAC9C,WAAWA,6BAAAA,CAAAA,CAAAA,OACJ,CAAC;QAAC;QAAQ;QAAM;QAAS;KAAO,EACpC,OAAO,CAAC,QACR,QAAQ,CAAC;IACZ,YAAYA,6BAAAA,CAAAA,CAAAA,OACL,CAAC;QACJ;QACA;QACA;QACA;QACA;KACD,EACA,OAAO,CAAC,gBACR,QAAQ,CACP;IAEJ,UAAUA,6BAAAA,CAAAA,CAAAA,MACD,GACN,QAAQ,GACR,QAAQ,GACR,QAAQ,CAAC;IACZ,QAAQC,AAAAA,IAAAA,mCAAAA,yBAAAA,AAAAA,IACL,QAAQ,GACR,QAAQ,CACP;AAEN;AAEO,MAAMiB,qBAAqB,CAChCf,OAEON,aAAgE;QACrE,MAAM;QACN,aACE;QACF,gBAAgB;QAChB,aAAaoB;QACbd;IACF;AAIK,MAAMgB,+BAA+BnB,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IACnD,MAAMC,AAAAA,IAAAA,mCAAAA,yBAAAA,AAAAA,IAA4B,QAAQ,CAAC;IAC3C,IAAIA,AAAAA,IAAAA,mCAAAA,yBAAAA,AAAAA,IAA4B,QAAQ,CAAC;AAC3C;AAMO,MAAMmB,0BAA0B,CACrCjB,OAEON,aAGL;QACA,MAAM;QACN,aACE;QACF,gBAAgB;QAChB,aAAasB;QACbhB;IACF;AAGK,MAAMkB,6BAA6BrB,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IACjD,QAAQC,AAAAA,IAAAA,mCAAAA,yBAAAA,AAAAA,IAA4B,QAAQ,CAC1C;IAEF,UAAUD,6BAAAA,CAAAA,CAAAA,MACD,GACN,OAAO,CAAC,KACR,QAAQ,GACR,QAAQ,CAAC;AACd;AAMO,MAAMsB,wBAAwB,CACnCnB,OAEON,aAAsE;QAC3E,MAAM;QACN,aAAa;QACb,aAAawB;QACblB;IACF;AAGK,MAAMoB,yBAAyBvB,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IAC7C,OAAOC,AAAAA,IAAAA,mCAAAA,yBAAAA,AAAAA,IACJ,QAAQ,GACR,QAAQ,CACP;IAEJ,WAAWD,6BAAAA,CAAAA,CAAAA,OACJ,CAAC;QAAC;QAAM;QAAQ;QAAQ;KAAQ,EACpC,QAAQ,GACR,QAAQ,CACP;IAEJ,UAAUA,6BAAAA,CAAAA,CAAAA,MACD,GACN,QAAQ,GACR,QAAQ,CAAC;IACZ,KAAKC,AAAAA,IAAAA,mCAAAA,yBAAAA,AAAAA,IACF,QAAQ,GACR,QAAQ,CACP;IAEJ,UAAUD,6BAAAA,CAAAA,CAAAA,MACD,GACN,OAAO,CAAC,KACR,QAAQ,CAAC;IACZ,QAAQA,6BAAAA,CAAAA,CAAAA,MACC,GACN,QAAQ,GACR,QAAQ,CACP;AAEN;AAWO,MAAMwB,oBAAoB,CAC/BrB,OAEON,aAA8D;QACnE,MAAM;QACN,aACE;QACF,aAAa0B;QACbpB;IACF;AAIK,MAAMsB,8BAA8BzB,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IAClD,QAAQC,AAAAA,IAAAA,mCAAAA,yBAAAA,AAAAA,IAA4B,QAAQ,CAAC;AAC/C;AAKO,MAAMyB,yBAAyB,CACpCvB,OAEON,aAGL;QACA,MAAM;QACN,aAAaa;QACb,gBAAgB;QAChB,aAAae;QACbtB;IACF;AAIK,MAAMwB,0BAA0B3B,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IAC9C,WAAWA,6BAAAA,CAAAA,CAAAA,MAAQ,GAAG,QAAQ,CAAC;IAC/B,SAASA,6BAAAA,CAAAA,CAAAA,MACA,GACN,QAAQ,CACP;IAEJ,QAAQA,6BAAAA,CAAAA,CAAAA,OAAS,GAAG,QAAQ,CAAC;AAC/B;AAOO,MAAM4B,qBAAqB,IACzB/B,aAAgE;QACrE,MAAM;QACN,aAAa;QACb,aAAa8B;QACb,MAAM,OAAOE;YACX,IAAI,AAAyB,aAAzB,OAAOA,OAAO,QAChB,MAAM,IAAIC,MACR,CAAC,wDAAwD,EAAE,OAAOD,OAAO,OAAO,EAAE,EAAEA,MAAM,OAAO,IAAI,gBAAgB;YAIzHE,IAAAA,uBAAAA,QAAAA,AAAAA,EAAS,wBACP,CAAC,QAAQ,EAAEF,MAAM,SAAS,CAAC,WAAW,EAAEA,MAAM,OAAO,CAAC,UAAU,EAAEA,MAAM,MAAM,EAAE;YAGlF,IAAI,CAACA,MAAM,MAAM,EACf,MAAM,IAAIC,MACR,CAAC,kBAAkB,EAAED,MAAM,OAAO,IAAI,eAAe,cAAc,EAAEA,MAAM,SAAS,CAAC,CAAC,CAAC;QAG7F;IACF"}
+{"version":3,"file":"device/index.js","sources":["webpack/runtime/define_property_getters","webpack/runtime/has_own_property","webpack/runtime/make_namespace_object","../../../src/device/index.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 { getMidsceneLocationSchema } from '@/common';\nimport type {\n  ActionScrollParam,\n  DeviceAction,\n  LocateResultElement,\n} from '@/types';\nimport type { IModelConfig } from '@midscene/shared/env';\nimport type { ElementNode } from '@midscene/shared/extractor';\nimport { getDebug } from '@midscene/shared/logger';\nimport { _keyDefinitions } from '@midscene/shared/us-keyboard-layout';\nimport { z } from 'zod';\nimport type { ElementCacheFeature, Rect, Size, UIContext } from '../types';\n\nexport abstract class AbstractInterface {\n  abstract interfaceType: string;\n\n  abstract screenshotBase64(): Promise<string>;\n  abstract size(): Promise<Size>;\n  abstract actionSpace(): DeviceAction[];\n\n  abstract cacheFeatureForRect?(\n    rect: Rect,\n    options?: {\n      targetDescription?: string;\n      modelConfig?: IModelConfig;\n    },\n  ): Promise<ElementCacheFeature>;\n  abstract rectMatchesCacheFeature?(\n    feature: ElementCacheFeature,\n  ): Promise<Rect>;\n\n  abstract destroy?(): Promise<void>;\n\n  abstract describe?(): string;\n  abstract beforeInvokeAction?(actionName: string, param: any): Promise<void>;\n  abstract afterInvokeAction?(actionName: string, param: any): Promise<void>;\n\n  // @deprecated do NOT extend this method\n  abstract getElementsNodeTree?: () => Promise<ElementNode>;\n\n  // @deprecated do NOT extend this method\n  abstract url?: () => string | Promise<string>;\n\n  // @deprecated do NOT extend this method\n  abstract evaluateJavaScript?<T = any>(script: string): Promise<T>;\n\n  // @deprecated do NOT extend this method\n  abstract getContext?(): Promise<UIContext>;\n}\n\n// Generic function to define actions with proper type inference\n// TRuntime allows specifying a different type for the runtime parameter (after location resolution)\n// TReturn allows specifying the return type of the action\nexport const defineAction = <\n  TSchema extends z.ZodType | undefined = undefined,\n  TRuntime = TSchema extends z.ZodType ? z.infer<TSchema> : undefined,\n  TReturn = any,\n>(\n  config: {\n    name: string;\n    description: string;\n    interfaceAlias?: string;\n    paramSchema?: TSchema;\n    call: (param: TRuntime) => Promise<TReturn> | TReturn;\n  } & Partial<\n    Omit<\n      DeviceAction<TRuntime, TReturn>,\n      'name' | 'description' | 'interfaceAlias' | 'paramSchema' | 'call'\n    >\n  >,\n): DeviceAction<TRuntime, TReturn> => {\n  return config as any; // Type assertion needed because schema validation type differs from runtime type\n};\n\n// Tap\nexport const actionTapParamSchema = z.object({\n  locate: getMidsceneLocationSchema().describe('The element to be tapped'),\n});\n// Override the inferred type to use LocateResultElement for the runtime locate field\nexport type ActionTapParam = {\n  locate: LocateResultElement;\n};\n\nexport const defineActionTap = (\n  call: (param: ActionTapParam) => Promise<void>,\n): DeviceAction<ActionTapParam> => {\n  return defineAction<typeof actionTapParamSchema, ActionTapParam>({\n    name: 'Tap',\n    description: 'Tap the element',\n    interfaceAlias: 'aiTap',\n    paramSchema: actionTapParamSchema,\n    call,\n  });\n};\n\n// RightClick\nexport const actionRightClickParamSchema = z.object({\n  locate: getMidsceneLocationSchema().describe(\n    'The element to be right clicked',\n  ),\n});\nexport type ActionRightClickParam = {\n  locate: LocateResultElement;\n};\n\nexport const defineActionRightClick = (\n  call: (param: ActionRightClickParam) => Promise<void>,\n): DeviceAction<ActionRightClickParam> => {\n  return defineAction<\n    typeof actionRightClickParamSchema,\n    ActionRightClickParam\n  >({\n    name: 'RightClick',\n    description: 'Right click the element',\n    interfaceAlias: 'aiRightClick',\n    paramSchema: actionRightClickParamSchema,\n    call,\n  });\n};\n\n// DoubleClick\nexport const actionDoubleClickParamSchema = z.object({\n  locate: getMidsceneLocationSchema().describe(\n    'The element to be double clicked',\n  ),\n});\nexport type ActionDoubleClickParam = {\n  locate: LocateResultElement;\n};\n\nexport const defineActionDoubleClick = (\n  call: (param: ActionDoubleClickParam) => Promise<void>,\n): DeviceAction<ActionDoubleClickParam> => {\n  return defineAction<\n    typeof actionDoubleClickParamSchema,\n    ActionDoubleClickParam\n  >({\n    name: 'DoubleClick',\n    description: 'Double click the element',\n    interfaceAlias: 'aiDoubleClick',\n    paramSchema: actionDoubleClickParamSchema,\n    call,\n  });\n};\n\n// Hover\nexport const actionHoverParamSchema = z.object({\n  locate: getMidsceneLocationSchema().describe('The element to be hovered'),\n});\nexport type ActionHoverParam = {\n  locate: LocateResultElement;\n};\n\nexport const defineActionHover = (\n  call: (param: ActionHoverParam) => Promise<void>,\n): DeviceAction<ActionHoverParam> => {\n  return defineAction<typeof actionHoverParamSchema, ActionHoverParam>({\n    name: 'Hover',\n    description: 'Move the mouse to the element',\n    interfaceAlias: 'aiHover',\n    paramSchema: actionHoverParamSchema,\n    call,\n  });\n};\n\n// Input\nconst inputLocateDescription =\n  'the position of the placeholder or text content in the target input field. If there is no content, locate the center of the input field.';\nexport const actionInputParamSchema = z.object({\n  value: z\n    .union([z.string(), z.number()])\n    .transform((val) => String(val))\n    .describe(\n      'The text to input. Provide the final content for replace/append modes, or an empty string when using clear mode to remove existing text.',\n    ),\n  locate: getMidsceneLocationSchema()\n    .describe(inputLocateDescription)\n    .optional(),\n  mode: z\n    .enum(['replace', 'clear', 'append'])\n    .default('replace')\n    .optional()\n    .describe(\n      'Input mode: \"replace\" (default) - clear the field and input the value; \"append\" - append the value to existing content; \"clear\" - clear the field without inputting new text.',\n    ),\n});\nexport type ActionInputParam = {\n  value: string;\n  locate?: LocateResultElement;\n  mode?: 'replace' | 'clear' | 'append';\n};\n\nexport const defineActionInput = (\n  call: (param: ActionInputParam) => Promise<void>,\n): DeviceAction<ActionInputParam> => {\n  return defineAction<typeof actionInputParamSchema, ActionInputParam>({\n    name: 'Input',\n    description: 'Input the value into the element',\n    interfaceAlias: 'aiInput',\n    paramSchema: actionInputParamSchema,\n    call,\n  });\n};\n\n// KeyboardPress\nexport const actionKeyboardPressParamSchema = z.object({\n  locate: getMidsceneLocationSchema()\n    .describe('The element to be clicked before pressing the key')\n    .optional(),\n  keyName: z\n    .string()\n    .describe(\n      \"The key to be pressed. Use '+' for key combinations, e.g., 'Control+A', 'Shift+Enter'\",\n    ),\n});\nexport type ActionKeyboardPressParam = {\n  locate?: LocateResultElement;\n  keyName: string;\n};\n\nexport const defineActionKeyboardPress = (\n  call: (param: ActionKeyboardPressParam) => Promise<void>,\n): DeviceAction<ActionKeyboardPressParam> => {\n  return defineAction<\n    typeof actionKeyboardPressParamSchema,\n    ActionKeyboardPressParam\n  >({\n    name: 'KeyboardPress',\n    description:\n      'Press a key or key combination, like \"Enter\", \"Tab\", \"Escape\", or \"Control+A\", \"Shift+Enter\". Do not use this to type text.',\n    interfaceAlias: 'aiKeyboardPress',\n    paramSchema: actionKeyboardPressParamSchema,\n    call,\n  });\n};\n\n// Scroll\nexport const actionScrollParamSchema = z.object({\n  scrollType: z\n    .enum([\n      'singleAction',\n      'scrollToBottom',\n      'scrollToTop',\n      'scrollToRight',\n      'scrollToLeft',\n    ])\n    .default('singleAction')\n    .describe(\n      'The scroll behavior: \"singleAction\" for a single scroll action, \"scrollToBottom\" for scrolling to the bottom, \"scrollToTop\" for scrolling to the top, \"scrollToRight\" for scrolling to the right, \"scrollToLeft\" for scrolling to the left',\n    ),\n  direction: z\n    .enum(['down', 'up', 'right', 'left'])\n    .default('down')\n    .describe(\n      'The direction to scroll. Only effective when scrollType is \"singleAction\".',\n    ),\n  distance: z\n    .number()\n    .nullable()\n    .optional()\n    .describe('The distance in pixels to scroll'),\n  locate: getMidsceneLocationSchema()\n    .optional()\n    .describe(\n      'The target element to be scrolled on, like \"the table\" or \"the list\" or \"the content area\" or \"the scrollable area\"',\n    ),\n});\n\nexport const defineActionScroll = (\n  call: (param: ActionScrollParam) => Promise<void>,\n): DeviceAction<ActionScrollParam> => {\n  return defineAction<typeof actionScrollParamSchema, ActionScrollParam>({\n    name: 'Scroll',\n    description:\n      'Scroll the page or an element. The direction to scroll, the scroll type, and the distance to scroll. The distance is the number of pixels to scroll. If not specified, use `down` direction, `once` scroll type, and `null` distance.',\n    interfaceAlias: 'aiScroll',\n    paramSchema: actionScrollParamSchema,\n    call,\n  });\n};\n\n// DragAndDrop\nexport const actionDragAndDropParamSchema = z.object({\n  from: getMidsceneLocationSchema().describe('The position to be dragged'),\n  to: getMidsceneLocationSchema().describe('The position to be dropped'),\n});\nexport type ActionDragAndDropParam = {\n  from: LocateResultElement;\n  to: LocateResultElement;\n};\n\nexport const defineActionDragAndDrop = (\n  call: (param: ActionDragAndDropParam) => Promise<void>,\n): DeviceAction<ActionDragAndDropParam> => {\n  return defineAction<\n    typeof actionDragAndDropParamSchema,\n    ActionDragAndDropParam\n  >({\n    name: 'DragAndDrop',\n    description:\n      'Drag and drop (hold the mouse or finger down and move the mouse) ',\n    interfaceAlias: 'aiDragAndDrop',\n    paramSchema: actionDragAndDropParamSchema,\n    call,\n  });\n};\n\nexport const ActionLongPressParamSchema = z.object({\n  locate: getMidsceneLocationSchema().describe(\n    'The element to be long pressed',\n  ),\n  duration: z\n    .number()\n    .default(500)\n    .optional()\n    .describe('Long press duration in milliseconds'),\n});\n\nexport type ActionLongPressParam = {\n  locate: LocateResultElement;\n  duration?: number;\n};\nexport const defineActionLongPress = (\n  call: (param: ActionLongPressParam) => Promise<void>,\n): DeviceAction<ActionLongPressParam> => {\n  return defineAction<typeof ActionLongPressParamSchema, ActionLongPressParam>({\n    name: 'LongPress',\n    description: 'Long press the element',\n    paramSchema: ActionLongPressParamSchema,\n    call,\n  });\n};\n\nexport const ActionSwipeParamSchema = z.object({\n  start: getMidsceneLocationSchema()\n    .optional()\n    .describe(\n      'Starting point of the swipe gesture, if not specified, the center of the page will be used',\n    ),\n  direction: z\n    .enum(['up', 'down', 'left', 'right'])\n    .optional()\n    .describe(\n      'The direction to swipe (required when using distance). The direction means the direction of the finger swipe.',\n    ),\n  distance: z\n    .number()\n    .optional()\n    .describe('The distance in pixels to swipe (mutually exclusive with end)'),\n  end: getMidsceneLocationSchema()\n    .optional()\n    .describe(\n      'Ending point of the swipe gesture (mutually exclusive with distance)',\n    ),\n  duration: z\n    .number()\n    .default(300)\n    .describe('Duration of the swipe gesture in milliseconds'),\n  repeat: z\n    .number()\n    .optional()\n    .describe(\n      'The number of times to repeat the swipe gesture. 1 for default, 0 for infinite (e.g. endless swipe until the end of the page)',\n    ),\n});\n\nexport type ActionSwipeParam = {\n  start?: LocateResultElement;\n  direction?: 'up' | 'down' | 'left' | 'right';\n  distance?: number;\n  end?: LocateResultElement;\n  duration?: number;\n  repeat?: number;\n};\n\nexport const defineActionSwipe = (\n  call: (param: ActionSwipeParam) => Promise<void>,\n): DeviceAction<ActionSwipeParam> => {\n  return defineAction<typeof ActionSwipeParamSchema, ActionSwipeParam>({\n    name: 'Swipe',\n    description:\n      'Perform a swipe gesture. You must specify either \"end\" (target location) or \"distance\" + \"direction\" - they are mutually exclusive. Use \"end\" for precise location-based swipes, or \"distance\" + \"direction\" for relative movement.',\n    paramSchema: ActionSwipeParamSchema,\n    call,\n  });\n};\n\n// ClearInput\nexport const actionClearInputParamSchema = z.object({\n  locate: getMidsceneLocationSchema().describe('The input field to be cleared'),\n});\nexport type ActionClearInputParam = {\n  locate: LocateResultElement;\n};\n\nexport const defineActionClearInput = (\n  call: (param: ActionClearInputParam) => Promise<void>,\n): DeviceAction<ActionClearInputParam> => {\n  return defineAction<\n    typeof actionClearInputParamSchema,\n    ActionClearInputParam\n  >({\n    name: 'ClearInput',\n    description: inputLocateDescription,\n    interfaceAlias: 'aiClearInput',\n    paramSchema: actionClearInputParamSchema,\n    call,\n  });\n};\n\n// Assert\nexport const actionAssertParamSchema = z.object({\n  condition: z.string().describe('The condition of the assertion'),\n  thought: z\n    .string()\n    .describe(\n      'The thought of the assertion, like \"I can see there are A, B, C elements on the page, which means ... , so the assertion is true\"',\n    ),\n  result: z.boolean().describe('The result of the assertion, true or false'),\n});\nexport type ActionAssertParam = {\n  condition: string;\n  thought: string;\n  result: boolean;\n};\n\nexport const defineActionAssert = (): DeviceAction<ActionAssertParam> => {\n  return defineAction<typeof actionAssertParamSchema, ActionAssertParam>({\n    name: 'Print_Assert_Result',\n    description: 'Print the result of the assertion',\n    paramSchema: actionAssertParamSchema,\n    call: async (param) => {\n      if (typeof param?.result !== 'boolean') {\n        throw new Error(\n          `The result of the assertion must be a boolean, but got: ${typeof param?.result}. ${param.thought || '(no thought)'}`,\n        );\n      }\n\n      getDebug('device:common-action')(\n        `Assert: ${param.condition}, Thought: ${param.thought}, Result: ${param.result}`,\n      );\n\n      if (!param.result) {\n        throw new Error(\n          `Assertion failed: ${param.thought || '(no thought)'} (Assertion = ${param.condition})`,\n        );\n      }\n    },\n  });\n};\n\nexport type { DeviceAction } from '../types';\nexport type {\n  AndroidDeviceOpt,\n  AndroidDeviceInputOpt,\n  IOSDeviceOpt,\n  IOSDeviceInputOpt,\n} from './device-options';\n"],"names":["__webpack_require__","definition","key","Object","obj","prop","Symbol","AbstractInterface","defineAction","config","actionTapParamSchema","z","getMidsceneLocationSchema","defineActionTap","call","actionRightClickParamSchema","defineActionRightClick","actionDoubleClickParamSchema","defineActionDoubleClick","actionHoverParamSchema","defineActionHover","inputLocateDescription","actionInputParamSchema","val","String","defineActionInput","actionKeyboardPressParamSchema","defineActionKeyboardPress","actionScrollParamSchema","defineActionScroll","actionDragAndDropParamSchema","defineActionDragAndDrop","ActionLongPressParamSchema","defineActionLongPress","ActionSwipeParamSchema","defineActionSwipe","actionClearInputParamSchema","defineActionClearInput","actionAssertParamSchema","defineActionAssert","param","Error","getDebug"],"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;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;ACOO,MAAeI;AAmCtB;AAKO,MAAMC,eAAe,CAK1BC,SAaOA;AAIF,MAAMC,uBAAuBC,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IAC3C,QAAQC,AAAAA,IAAAA,mCAAAA,yBAAAA,AAAAA,IAA4B,QAAQ,CAAC;AAC/C;AAMO,MAAMC,kBAAkB,CAC7BC,OAEON,aAA0D;QAC/D,MAAM;QACN,aAAa;QACb,gBAAgB;QAChB,aAAaE;QACbI;IACF;AAIK,MAAMC,8BAA8BJ,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IAClD,QAAQC,AAAAA,IAAAA,mCAAAA,yBAAAA,AAAAA,IAA4B,QAAQ,CAC1C;AAEJ;AAKO,MAAMI,yBAAyB,CACpCF,OAEON,aAGL;QACA,MAAM;QACN,aAAa;QACb,gBAAgB;QAChB,aAAaO;QACbD;IACF;AAIK,MAAMG,+BAA+BN,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IACnD,QAAQC,AAAAA,IAAAA,mCAAAA,yBAAAA,AAAAA,IAA4B,QAAQ,CAC1C;AAEJ;AAKO,MAAMM,0BAA0B,CACrCJ,OAEON,aAGL;QACA,MAAM;QACN,aAAa;QACb,gBAAgB;QAChB,aAAaS;QACbH;IACF;AAIK,MAAMK,yBAAyBR,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IAC7C,QAAQC,AAAAA,IAAAA,mCAAAA,yBAAAA,AAAAA,IAA4B,QAAQ,CAAC;AAC/C;AAKO,MAAMQ,oBAAoB,CAC/BN,OAEON,aAA8D;QACnE,MAAM;QACN,aAAa;QACb,gBAAgB;QAChB,aAAaW;QACbL;IACF;AAIF,MAAMO,yBACJ;AACK,MAAMC,yBAAyBX,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IAC7C,OAAOA,6BAAAA,CAAAA,CAAAA,KACC,CAAC;QAACA,6BAAAA,CAAAA,CAAAA,MAAQ;QAAIA,6BAAAA,CAAAA,CAAAA,MAAQ;KAAG,EAC9B,SAAS,CAAC,CAACY,MAAQC,OAAOD,MAC1B,QAAQ,CACP;IAEJ,QAAQX,AAAAA,IAAAA,mCAAAA,yBAAAA,AAAAA,IACL,QAAQ,CAACS,wBACT,QAAQ;IACX,MAAMV,6BAAAA,CAAAA,CAAAA,OACC,CAAC;QAAC;QAAW;QAAS;KAAS,EACnC,OAAO,CAAC,WACR,QAAQ,GACR,QAAQ,CACP;AAEN;AAOO,MAAMc,oBAAoB,CAC/BX,OAEON,aAA8D;QACnE,MAAM;QACN,aAAa;QACb,gBAAgB;QAChB,aAAac;QACbR;IACF;AAIK,MAAMY,iCAAiCf,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IACrD,QAAQC,AAAAA,IAAAA,mCAAAA,yBAAAA,AAAAA,IACL,QAAQ,CAAC,qDACT,QAAQ;IACX,SAASD,6BAAAA,CAAAA,CAAAA,MACA,GACN,QAAQ,CACP;AAEN;AAMO,MAAMgB,4BAA4B,CACvCb,OAEON,aAGL;QACA,MAAM;QACN,aACE;QACF,gBAAgB;QAChB,aAAakB;QACbZ;IACF;AAIK,MAAMc,0BAA0BjB,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IAC9C,YAAYA,6BAAAA,CAAAA,CAAAA,OACL,CAAC;QACJ;QACA;QACA;QACA;QACA;KACD,EACA,OAAO,CAAC,gBACR,QAAQ,CACP;IAEJ,WAAWA,6BAAAA,CAAAA,CAAAA,OACJ,CAAC;QAAC;QAAQ;QAAM;QAAS;KAAO,EACpC,OAAO,CAAC,QACR,QAAQ,CACP;IAEJ,UAAUA,6BAAAA,CAAAA,CAAAA,MACD,GACN,QAAQ,GACR,QAAQ,GACR,QAAQ,CAAC;IACZ,QAAQC,AAAAA,IAAAA,mCAAAA,yBAAAA,AAAAA,IACL,QAAQ,GACR,QAAQ,CACP;AAEN;AAEO,MAAMiB,qBAAqB,CAChCf,OAEON,aAAgE;QACrE,MAAM;QACN,aACE;QACF,gBAAgB;QAChB,aAAaoB;QACbd;IACF;AAIK,MAAMgB,+BAA+BnB,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IACnD,MAAMC,AAAAA,IAAAA,mCAAAA,yBAAAA,AAAAA,IAA4B,QAAQ,CAAC;IAC3C,IAAIA,AAAAA,IAAAA,mCAAAA,yBAAAA,AAAAA,IAA4B,QAAQ,CAAC;AAC3C;AAMO,MAAMmB,0BAA0B,CACrCjB,OAEON,aAGL;QACA,MAAM;QACN,aACE;QACF,gBAAgB;QAChB,aAAasB;QACbhB;IACF;AAGK,MAAMkB,6BAA6BrB,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IACjD,QAAQC,AAAAA,IAAAA,mCAAAA,yBAAAA,AAAAA,IAA4B,QAAQ,CAC1C;IAEF,UAAUD,6BAAAA,CAAAA,CAAAA,MACD,GACN,OAAO,CAAC,KACR,QAAQ,GACR,QAAQ,CAAC;AACd;AAMO,MAAMsB,wBAAwB,CACnCnB,OAEON,aAAsE;QAC3E,MAAM;QACN,aAAa;QACb,aAAawB;QACblB;IACF;AAGK,MAAMoB,yBAAyBvB,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IAC7C,OAAOC,AAAAA,IAAAA,mCAAAA,yBAAAA,AAAAA,IACJ,QAAQ,GACR,QAAQ,CACP;IAEJ,WAAWD,6BAAAA,CAAAA,CAAAA,OACJ,CAAC;QAAC;QAAM;QAAQ;QAAQ;KAAQ,EACpC,QAAQ,GACR,QAAQ,CACP;IAEJ,UAAUA,6BAAAA,CAAAA,CAAAA,MACD,GACN,QAAQ,GACR,QAAQ,CAAC;IACZ,KAAKC,AAAAA,IAAAA,mCAAAA,yBAAAA,AAAAA,IACF,QAAQ,GACR,QAAQ,CACP;IAEJ,UAAUD,6BAAAA,CAAAA,CAAAA,MACD,GACN,OAAO,CAAC,KACR,QAAQ,CAAC;IACZ,QAAQA,6BAAAA,CAAAA,CAAAA,MACC,GACN,QAAQ,GACR,QAAQ,CACP;AAEN;AAWO,MAAMwB,oBAAoB,CAC/BrB,OAEON,aAA8D;QACnE,MAAM;QACN,aACE;QACF,aAAa0B;QACbpB;IACF;AAIK,MAAMsB,8BAA8BzB,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IAClD,QAAQC,AAAAA,IAAAA,mCAAAA,yBAAAA,AAAAA,IAA4B,QAAQ,CAAC;AAC/C;AAKO,MAAMyB,yBAAyB,CACpCvB,OAEON,aAGL;QACA,MAAM;QACN,aAAaa;QACb,gBAAgB;QAChB,aAAae;QACbtB;IACF;AAIK,MAAMwB,0BAA0B3B,6BAAAA,CAAAA,CAAAA,MAAQ,CAAC;IAC9C,WAAWA,6BAAAA,CAAAA,CAAAA,MAAQ,GAAG,QAAQ,CAAC;IAC/B,SAASA,6BAAAA,CAAAA,CAAAA,MACA,GACN,QAAQ,CACP;IAEJ,QAAQA,6BAAAA,CAAAA,CAAAA,OAAS,GAAG,QAAQ,CAAC;AAC/B;AAOO,MAAM4B,qBAAqB,IACzB/B,aAAgE;QACrE,MAAM;QACN,aAAa;QACb,aAAa8B;QACb,MAAM,OAAOE;YACX,IAAI,AAAyB,aAAzB,OAAOA,OAAO,QAChB,MAAM,IAAIC,MACR,CAAC,wDAAwD,EAAE,OAAOD,OAAO,OAAO,EAAE,EAAEA,MAAM,OAAO,IAAI,gBAAgB;YAIzHE,IAAAA,uBAAAA,QAAAA,AAAAA,EAAS,wBACP,CAAC,QAAQ,EAAEF,MAAM,SAAS,CAAC,WAAW,EAAEA,MAAM,OAAO,CAAC,UAAU,EAAEA,MAAM,MAAM,EAAE;YAGlF,IAAI,CAACA,MAAM,MAAM,EACf,MAAM,IAAIC,MACR,CAAC,kBAAkB,EAAED,MAAM,OAAO,IAAI,eAAe,cAAc,EAAEA,MAAM,SAAS,CAAC,CAAC,CAAC;QAG7F;IACF"}