@aiscene/core 1.7.0

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 (279) hide show
  1. package/LICENSE +21 -0
  2. package/README.md +9 -0
  3. package/dist/es/agent/agent.mjs +749 -0
  4. package/dist/es/agent/agent.mjs.map +1 -0
  5. package/dist/es/agent/common.mjs +0 -0
  6. package/dist/es/agent/execution-session.mjs +41 -0
  7. package/dist/es/agent/execution-session.mjs.map +1 -0
  8. package/dist/es/agent/index.mjs +6 -0
  9. package/dist/es/agent/task-builder.mjs +332 -0
  10. package/dist/es/agent/task-builder.mjs.map +1 -0
  11. package/dist/es/agent/task-cache.mjs +214 -0
  12. package/dist/es/agent/task-cache.mjs.map +1 -0
  13. package/dist/es/agent/tasks.mjs +426 -0
  14. package/dist/es/agent/tasks.mjs.map +1 -0
  15. package/dist/es/agent/ui-utils.mjs +91 -0
  16. package/dist/es/agent/ui-utils.mjs.map +1 -0
  17. package/dist/es/agent/utils.mjs +198 -0
  18. package/dist/es/agent/utils.mjs.map +1 -0
  19. package/dist/es/ai-model/auto-glm/actions.mjs +237 -0
  20. package/dist/es/ai-model/auto-glm/actions.mjs.map +1 -0
  21. package/dist/es/ai-model/auto-glm/index.mjs +6 -0
  22. package/dist/es/ai-model/auto-glm/parser.mjs +239 -0
  23. package/dist/es/ai-model/auto-glm/parser.mjs.map +1 -0
  24. package/dist/es/ai-model/auto-glm/planning.mjs +71 -0
  25. package/dist/es/ai-model/auto-glm/planning.mjs.map +1 -0
  26. package/dist/es/ai-model/auto-glm/prompt.mjs +222 -0
  27. package/dist/es/ai-model/auto-glm/prompt.mjs.map +1 -0
  28. package/dist/es/ai-model/auto-glm/util.mjs +9 -0
  29. package/dist/es/ai-model/auto-glm/util.mjs.map +1 -0
  30. package/dist/es/ai-model/conversation-history.mjs +195 -0
  31. package/dist/es/ai-model/conversation-history.mjs.map +1 -0
  32. package/dist/es/ai-model/index.mjs +11 -0
  33. package/dist/es/ai-model/inspect.mjs +394 -0
  34. package/dist/es/ai-model/inspect.mjs.map +1 -0
  35. package/dist/es/ai-model/llm-planning.mjs +233 -0
  36. package/dist/es/ai-model/llm-planning.mjs.map +1 -0
  37. package/dist/es/ai-model/prompt/common.mjs +7 -0
  38. package/dist/es/ai-model/prompt/common.mjs.map +1 -0
  39. package/dist/es/ai-model/prompt/describe.mjs +66 -0
  40. package/dist/es/ai-model/prompt/describe.mjs.map +1 -0
  41. package/dist/es/ai-model/prompt/extraction.mjs +129 -0
  42. package/dist/es/ai-model/prompt/extraction.mjs.map +1 -0
  43. package/dist/es/ai-model/prompt/llm-locator.mjs +51 -0
  44. package/dist/es/ai-model/prompt/llm-locator.mjs.map +1 -0
  45. package/dist/es/ai-model/prompt/llm-planning.mjs +568 -0
  46. package/dist/es/ai-model/prompt/llm-planning.mjs.map +1 -0
  47. package/dist/es/ai-model/prompt/llm-section-locator.mjs +44 -0
  48. package/dist/es/ai-model/prompt/llm-section-locator.mjs.map +1 -0
  49. package/dist/es/ai-model/prompt/order-sensitive-judge.mjs +35 -0
  50. package/dist/es/ai-model/prompt/order-sensitive-judge.mjs.map +1 -0
  51. package/dist/es/ai-model/prompt/playwright-generator.mjs +117 -0
  52. package/dist/es/ai-model/prompt/playwright-generator.mjs.map +1 -0
  53. package/dist/es/ai-model/prompt/ui-tars-planning.mjs +36 -0
  54. package/dist/es/ai-model/prompt/ui-tars-planning.mjs.map +1 -0
  55. package/dist/es/ai-model/prompt/util.mjs +59 -0
  56. package/dist/es/ai-model/prompt/util.mjs.map +1 -0
  57. package/dist/es/ai-model/prompt/yaml-generator.mjs +219 -0
  58. package/dist/es/ai-model/prompt/yaml-generator.mjs.map +1 -0
  59. package/dist/es/ai-model/service-caller/codex-app-server.mjs +575 -0
  60. package/dist/es/ai-model/service-caller/codex-app-server.mjs.map +1 -0
  61. package/dist/es/ai-model/service-caller/image-detail.mjs +6 -0
  62. package/dist/es/ai-model/service-caller/image-detail.mjs.map +1 -0
  63. package/dist/es/ai-model/service-caller/index.mjs +473 -0
  64. package/dist/es/ai-model/service-caller/index.mjs.map +1 -0
  65. package/dist/es/ai-model/ui-tars-planning.mjs +249 -0
  66. package/dist/es/ai-model/ui-tars-planning.mjs.map +1 -0
  67. package/dist/es/common.mjs +371 -0
  68. package/dist/es/common.mjs.map +1 -0
  69. package/dist/es/device/device-options.mjs +0 -0
  70. package/dist/es/device/index.mjs +341 -0
  71. package/dist/es/device/index.mjs.map +1 -0
  72. package/dist/es/dump/html-utils.mjs +284 -0
  73. package/dist/es/dump/html-utils.mjs.map +1 -0
  74. package/dist/es/dump/image-restoration.mjs +43 -0
  75. package/dist/es/dump/image-restoration.mjs.map +1 -0
  76. package/dist/es/dump/index.mjs +3 -0
  77. package/dist/es/index.mjs +15 -0
  78. package/dist/es/index.mjs.map +1 -0
  79. package/dist/es/report-generator.mjs +162 -0
  80. package/dist/es/report-generator.mjs.map +1 -0
  81. package/dist/es/report.mjs +137 -0
  82. package/dist/es/report.mjs.map +1 -0
  83. package/dist/es/screenshot-item.mjs +105 -0
  84. package/dist/es/screenshot-item.mjs.map +1 -0
  85. package/dist/es/service/index.mjs +274 -0
  86. package/dist/es/service/index.mjs.map +1 -0
  87. package/dist/es/service/utils.mjs +15 -0
  88. package/dist/es/service/utils.mjs.map +1 -0
  89. package/dist/es/skill/index.mjs +38 -0
  90. package/dist/es/skill/index.mjs.map +1 -0
  91. package/dist/es/task-runner.mjs +263 -0
  92. package/dist/es/task-runner.mjs.map +1 -0
  93. package/dist/es/task-timing.mjs +12 -0
  94. package/dist/es/task-timing.mjs.map +1 -0
  95. package/dist/es/tree.mjs +13 -0
  96. package/dist/es/tree.mjs.map +1 -0
  97. package/dist/es/types.mjs +199 -0
  98. package/dist/es/types.mjs.map +1 -0
  99. package/dist/es/utils.mjs +229 -0
  100. package/dist/es/utils.mjs.map +1 -0
  101. package/dist/es/yaml/builder.mjs +13 -0
  102. package/dist/es/yaml/builder.mjs.map +1 -0
  103. package/dist/es/yaml/index.mjs +4 -0
  104. package/dist/es/yaml/player.mjs +434 -0
  105. package/dist/es/yaml/player.mjs.map +1 -0
  106. package/dist/es/yaml/utils.mjs +102 -0
  107. package/dist/es/yaml/utils.mjs.map +1 -0
  108. package/dist/es/yaml.mjs +0 -0
  109. package/dist/lib/agent/agent.js +797 -0
  110. package/dist/lib/agent/agent.js.map +1 -0
  111. package/dist/lib/agent/common.js +5 -0
  112. package/dist/lib/agent/execution-session.js +75 -0
  113. package/dist/lib/agent/execution-session.js.map +1 -0
  114. package/dist/lib/agent/index.js +81 -0
  115. package/dist/lib/agent/index.js.map +1 -0
  116. package/dist/lib/agent/task-builder.js +369 -0
  117. package/dist/lib/agent/task-builder.js.map +1 -0
  118. package/dist/lib/agent/task-cache.js +266 -0
  119. package/dist/lib/agent/task-cache.js.map +1 -0
  120. package/dist/lib/agent/tasks.js +469 -0
  121. package/dist/lib/agent/tasks.js.map +1 -0
  122. package/dist/lib/agent/ui-utils.js +143 -0
  123. package/dist/lib/agent/ui-utils.js.map +1 -0
  124. package/dist/lib/agent/utils.js +275 -0
  125. package/dist/lib/agent/utils.js.map +1 -0
  126. package/dist/lib/ai-model/auto-glm/actions.js +271 -0
  127. package/dist/lib/ai-model/auto-glm/actions.js.map +1 -0
  128. package/dist/lib/ai-model/auto-glm/index.js +66 -0
  129. package/dist/lib/ai-model/auto-glm/index.js.map +1 -0
  130. package/dist/lib/ai-model/auto-glm/parser.js +282 -0
  131. package/dist/lib/ai-model/auto-glm/parser.js.map +1 -0
  132. package/dist/lib/ai-model/auto-glm/planning.js +105 -0
  133. package/dist/lib/ai-model/auto-glm/planning.js.map +1 -0
  134. package/dist/lib/ai-model/auto-glm/prompt.js +259 -0
  135. package/dist/lib/ai-model/auto-glm/prompt.js.map +1 -0
  136. package/dist/lib/ai-model/auto-glm/util.js +46 -0
  137. package/dist/lib/ai-model/auto-glm/util.js.map +1 -0
  138. package/dist/lib/ai-model/conversation-history.js +229 -0
  139. package/dist/lib/ai-model/conversation-history.js.map +1 -0
  140. package/dist/lib/ai-model/index.js +125 -0
  141. package/dist/lib/ai-model/index.js.map +1 -0
  142. package/dist/lib/ai-model/inspect.js +440 -0
  143. package/dist/lib/ai-model/inspect.js.map +1 -0
  144. package/dist/lib/ai-model/llm-planning.js +270 -0
  145. package/dist/lib/ai-model/llm-planning.js.map +1 -0
  146. package/dist/lib/ai-model/prompt/common.js +41 -0
  147. package/dist/lib/ai-model/prompt/common.js.map +1 -0
  148. package/dist/lib/ai-model/prompt/describe.js +100 -0
  149. package/dist/lib/ai-model/prompt/describe.js.map +1 -0
  150. package/dist/lib/ai-model/prompt/extraction.js +169 -0
  151. package/dist/lib/ai-model/prompt/extraction.js.map +1 -0
  152. package/dist/lib/ai-model/prompt/llm-locator.js +88 -0
  153. package/dist/lib/ai-model/prompt/llm-locator.js.map +1 -0
  154. package/dist/lib/ai-model/prompt/llm-planning.js +605 -0
  155. package/dist/lib/ai-model/prompt/llm-planning.js.map +1 -0
  156. package/dist/lib/ai-model/prompt/llm-section-locator.js +81 -0
  157. package/dist/lib/ai-model/prompt/llm-section-locator.js.map +1 -0
  158. package/dist/lib/ai-model/prompt/order-sensitive-judge.js +72 -0
  159. package/dist/lib/ai-model/prompt/order-sensitive-judge.js.map +1 -0
  160. package/dist/lib/ai-model/prompt/playwright-generator.js +178 -0
  161. package/dist/lib/ai-model/prompt/playwright-generator.js.map +1 -0
  162. package/dist/lib/ai-model/prompt/ui-tars-planning.js +73 -0
  163. package/dist/lib/ai-model/prompt/ui-tars-planning.js.map +1 -0
  164. package/dist/lib/ai-model/prompt/util.js +105 -0
  165. package/dist/lib/ai-model/prompt/util.js.map +1 -0
  166. package/dist/lib/ai-model/prompt/yaml-generator.js +280 -0
  167. package/dist/lib/ai-model/prompt/yaml-generator.js.map +1 -0
  168. package/dist/lib/ai-model/service-caller/codex-app-server.js +624 -0
  169. package/dist/lib/ai-model/service-caller/codex-app-server.js.map +1 -0
  170. package/dist/lib/ai-model/service-caller/image-detail.js +40 -0
  171. package/dist/lib/ai-model/service-caller/image-detail.js.map +1 -0
  172. package/dist/lib/ai-model/service-caller/index.js +538 -0
  173. package/dist/lib/ai-model/service-caller/index.js.map +1 -0
  174. package/dist/lib/ai-model/ui-tars-planning.js +283 -0
  175. package/dist/lib/ai-model/ui-tars-planning.js.map +1 -0
  176. package/dist/lib/common.js +480 -0
  177. package/dist/lib/common.js.map +1 -0
  178. package/dist/lib/device/device-options.js +20 -0
  179. package/dist/lib/device/device-options.js.map +1 -0
  180. package/dist/lib/device/index.js +468 -0
  181. package/dist/lib/device/index.js.map +1 -0
  182. package/dist/lib/dump/html-utils.js +357 -0
  183. package/dist/lib/dump/html-utils.js.map +1 -0
  184. package/dist/lib/dump/image-restoration.js +77 -0
  185. package/dist/lib/dump/image-restoration.js.map +1 -0
  186. package/dist/lib/dump/index.js +60 -0
  187. package/dist/lib/dump/index.js.map +1 -0
  188. package/dist/lib/index.js +146 -0
  189. package/dist/lib/index.js.map +1 -0
  190. package/dist/lib/report-generator.js +200 -0
  191. package/dist/lib/report-generator.js.map +1 -0
  192. package/dist/lib/report.js +171 -0
  193. package/dist/lib/report.js.map +1 -0
  194. package/dist/lib/screenshot-item.js +139 -0
  195. package/dist/lib/screenshot-item.js.map +1 -0
  196. package/dist/lib/service/index.js +308 -0
  197. package/dist/lib/service/index.js.map +1 -0
  198. package/dist/lib/service/utils.js +49 -0
  199. package/dist/lib/service/utils.js.map +1 -0
  200. package/dist/lib/skill/index.js +72 -0
  201. package/dist/lib/skill/index.js.map +1 -0
  202. package/dist/lib/task-runner.js +300 -0
  203. package/dist/lib/task-runner.js.map +1 -0
  204. package/dist/lib/task-timing.js +46 -0
  205. package/dist/lib/task-timing.js.map +1 -0
  206. package/dist/lib/tree.js +53 -0
  207. package/dist/lib/tree.js.map +1 -0
  208. package/dist/lib/types.js +288 -0
  209. package/dist/lib/types.js.map +1 -0
  210. package/dist/lib/utils.js +308 -0
  211. package/dist/lib/utils.js.map +1 -0
  212. package/dist/lib/yaml/builder.js +57 -0
  213. package/dist/lib/yaml/builder.js.map +1 -0
  214. package/dist/lib/yaml/index.js +81 -0
  215. package/dist/lib/yaml/index.js.map +1 -0
  216. package/dist/lib/yaml/player.js +468 -0
  217. package/dist/lib/yaml/player.js.map +1 -0
  218. package/dist/lib/yaml/utils.js +155 -0
  219. package/dist/lib/yaml/utils.js.map +1 -0
  220. package/dist/lib/yaml.js +20 -0
  221. package/dist/lib/yaml.js.map +1 -0
  222. package/dist/types/agent/agent.d.ts +205 -0
  223. package/dist/types/agent/common.d.ts +0 -0
  224. package/dist/types/agent/execution-session.d.ts +36 -0
  225. package/dist/types/agent/index.d.ts +10 -0
  226. package/dist/types/agent/task-builder.d.ts +34 -0
  227. package/dist/types/agent/task-cache.d.ts +49 -0
  228. package/dist/types/agent/tasks.d.ts +69 -0
  229. package/dist/types/agent/ui-utils.d.ts +14 -0
  230. package/dist/types/agent/utils.d.ts +31 -0
  231. package/dist/types/ai-model/auto-glm/actions.d.ts +78 -0
  232. package/dist/types/ai-model/auto-glm/index.d.ts +6 -0
  233. package/dist/types/ai-model/auto-glm/parser.d.ts +18 -0
  234. package/dist/types/ai-model/auto-glm/planning.d.ts +12 -0
  235. package/dist/types/ai-model/auto-glm/prompt.d.ts +27 -0
  236. package/dist/types/ai-model/auto-glm/util.d.ts +13 -0
  237. package/dist/types/ai-model/conversation-history.d.ts +105 -0
  238. package/dist/types/ai-model/index.d.ts +14 -0
  239. package/dist/types/ai-model/inspect.d.ts +67 -0
  240. package/dist/types/ai-model/llm-planning.d.ts +19 -0
  241. package/dist/types/ai-model/prompt/common.d.ts +2 -0
  242. package/dist/types/ai-model/prompt/describe.d.ts +1 -0
  243. package/dist/types/ai-model/prompt/extraction.d.ts +7 -0
  244. package/dist/types/ai-model/prompt/llm-locator.d.ts +3 -0
  245. package/dist/types/ai-model/prompt/llm-planning.d.ts +10 -0
  246. package/dist/types/ai-model/prompt/llm-section-locator.d.ts +3 -0
  247. package/dist/types/ai-model/prompt/order-sensitive-judge.d.ts +2 -0
  248. package/dist/types/ai-model/prompt/playwright-generator.d.ts +26 -0
  249. package/dist/types/ai-model/prompt/ui-tars-planning.d.ts +2 -0
  250. package/dist/types/ai-model/prompt/util.d.ts +33 -0
  251. package/dist/types/ai-model/prompt/yaml-generator.d.ts +100 -0
  252. package/dist/types/ai-model/service-caller/codex-app-server.d.ts +42 -0
  253. package/dist/types/ai-model/service-caller/image-detail.d.ts +2 -0
  254. package/dist/types/ai-model/service-caller/index.d.ts +49 -0
  255. package/dist/types/ai-model/ui-tars-planning.d.ts +72 -0
  256. package/dist/types/common.d.ts +288 -0
  257. package/dist/types/device/device-options.d.ts +142 -0
  258. package/dist/types/device/index.d.ts +2528 -0
  259. package/dist/types/dump/html-utils.d.ts +63 -0
  260. package/dist/types/dump/image-restoration.d.ts +6 -0
  261. package/dist/types/dump/index.d.ts +5 -0
  262. package/dist/types/index.d.ts +17 -0
  263. package/dist/types/report-generator.d.ts +66 -0
  264. package/dist/types/report.d.ts +22 -0
  265. package/dist/types/screenshot-item.d.ts +66 -0
  266. package/dist/types/service/index.d.ts +24 -0
  267. package/dist/types/service/utils.d.ts +2 -0
  268. package/dist/types/skill/index.d.ts +25 -0
  269. package/dist/types/task-runner.d.ts +50 -0
  270. package/dist/types/task-timing.d.ts +8 -0
  271. package/dist/types/tree.d.ts +4 -0
  272. package/dist/types/types.d.ts +669 -0
  273. package/dist/types/utils.d.ts +40 -0
  274. package/dist/types/yaml/builder.d.ts +2 -0
  275. package/dist/types/yaml/index.d.ts +4 -0
  276. package/dist/types/yaml/player.d.ts +34 -0
  277. package/dist/types/yaml/utils.d.ts +9 -0
  278. package/dist/types/yaml.d.ts +217 -0
  279. package/package.json +130 -0
@@ -0,0 +1,797 @@
1
+ "use strict";
2
+ var __webpack_require__ = {};
3
+ (()=>{
4
+ __webpack_require__.n = (module)=>{
5
+ var getter = module && module.__esModule ? ()=>module['default'] : ()=>module;
6
+ __webpack_require__.d(getter, {
7
+ a: getter
8
+ });
9
+ return getter;
10
+ };
11
+ })();
12
+ (()=>{
13
+ __webpack_require__.d = (exports1, definition)=>{
14
+ for(var key in definition)if (__webpack_require__.o(definition, key) && !__webpack_require__.o(exports1, key)) Object.defineProperty(exports1, key, {
15
+ enumerable: true,
16
+ get: definition[key]
17
+ });
18
+ };
19
+ })();
20
+ (()=>{
21
+ __webpack_require__.o = (obj, prop)=>Object.prototype.hasOwnProperty.call(obj, prop);
22
+ })();
23
+ (()=>{
24
+ __webpack_require__.r = (exports1)=>{
25
+ if ('undefined' != typeof Symbol && Symbol.toStringTag) Object.defineProperty(exports1, Symbol.toStringTag, {
26
+ value: 'Module'
27
+ });
28
+ Object.defineProperty(exports1, '__esModule', {
29
+ value: true
30
+ });
31
+ };
32
+ })();
33
+ var __webpack_exports__ = {};
34
+ __webpack_require__.r(__webpack_exports__);
35
+ __webpack_require__.d(__webpack_exports__, {
36
+ Agent: ()=>Agent,
37
+ createAgent: ()=>createAgent
38
+ });
39
+ const external_screenshot_item_js_namespaceObject = require("../screenshot-item.js");
40
+ const index_js_namespaceObject = require("../service/index.js");
41
+ var index_js_default = /*#__PURE__*/ __webpack_require__.n(index_js_namespaceObject);
42
+ const external_types_js_namespaceObject = require("../types.js");
43
+ const util_js_namespaceObject = require("../ai-model/auto-glm/util.js");
44
+ const external_js_yaml_namespaceObject = require("js-yaml");
45
+ var external_js_yaml_default = /*#__PURE__*/ __webpack_require__.n(external_js_yaml_namespaceObject);
46
+ const external_report_generator_js_namespaceObject = require("../report-generator.js");
47
+ const external_utils_js_namespaceObject = require("../utils.js");
48
+ const external_yaml_index_js_namespaceObject = require("../yaml/index.js");
49
+ const external_node_fs_namespaceObject = require("node:fs");
50
+ const external_node_path_namespaceObject = require("node:path");
51
+ const env_namespaceObject = require("@midscene/shared/env");
52
+ const logger_namespaceObject = require("@midscene/shared/logger");
53
+ const utils_namespaceObject = require("@midscene/shared/utils");
54
+ const external_device_index_js_namespaceObject = require("../device/index.js");
55
+ const external_task_cache_js_namespaceObject = require("./task-cache.js");
56
+ const external_tasks_js_namespaceObject = require("./tasks.js");
57
+ const external_ui_utils_js_namespaceObject = require("./ui-utils.js");
58
+ const external_utils_js_namespaceObject_1 = require("./utils.js");
59
+ function _define_property(obj, key, value) {
60
+ if (key in obj) Object.defineProperty(obj, key, {
61
+ value: value,
62
+ enumerable: true,
63
+ configurable: true,
64
+ writable: true
65
+ });
66
+ else obj[key] = value;
67
+ return obj;
68
+ }
69
+ const debug = (0, logger_namespaceObject.getDebug)('agent');
70
+ const distanceOfTwoPoints = (p1, p2)=>{
71
+ const [x1, y1] = p1;
72
+ const [x2, y2] = p2;
73
+ return Math.round(Math.sqrt((x1 - x2) ** 2 + (y1 - y2) ** 2));
74
+ };
75
+ const includedInRect = (point, rect)=>{
76
+ const [x, y] = point;
77
+ const { left, top, width, height } = rect;
78
+ return x >= left && x <= left + width && y >= top && y <= top + height;
79
+ };
80
+ const defaultServiceExtractOption = {
81
+ domIncluded: false,
82
+ screenshotIncluded: true
83
+ };
84
+ const CACHE_STRATEGIES = [
85
+ 'read-only',
86
+ 'read-write',
87
+ 'write-only'
88
+ ];
89
+ const isValidCacheStrategy = (strategy)=>CACHE_STRATEGIES.some((value)=>value === strategy);
90
+ const CACHE_STRATEGY_VALUES = CACHE_STRATEGIES.map((value)=>`"${value}"`).join(', ');
91
+ const legacyScrollTypeMap = {
92
+ once: 'singleAction',
93
+ untilBottom: 'scrollToBottom',
94
+ untilTop: 'scrollToTop',
95
+ untilRight: 'scrollToRight',
96
+ untilLeft: 'scrollToLeft'
97
+ };
98
+ const normalizeScrollType = (scrollType)=>{
99
+ if (!scrollType) return scrollType;
100
+ if (scrollType in legacyScrollTypeMap) return legacyScrollTypeMap[scrollType];
101
+ return scrollType;
102
+ };
103
+ const defaultReplanningCycleLimit = 20;
104
+ const defaultVlmUiTarsReplanningCycleLimit = 40;
105
+ const defaultAutoGlmReplanningCycleLimit = 100;
106
+ class Agent {
107
+ get onDumpUpdate() {
108
+ return this.dumpUpdateListeners[0];
109
+ }
110
+ set onDumpUpdate(callback) {
111
+ this.dumpUpdateListeners = [];
112
+ if (callback) this.dumpUpdateListeners.push(callback);
113
+ }
114
+ get aiActContext() {
115
+ return this.opts.aiActContext ?? this.opts.aiActionContext;
116
+ }
117
+ get page() {
118
+ return this.interface;
119
+ }
120
+ ensureVLModelWarning() {
121
+ if (!this.hasWarnedNonVLModel && 'puppeteer' !== this.interface.interfaceType && 'playwright' !== this.interface.interfaceType && 'static' !== this.interface.interfaceType && 'chrome-extension-proxy' !== this.interface.interfaceType && 'page-over-chrome-extension-bridge' !== this.interface.interfaceType) {
122
+ this.modelConfigManager.throwErrorIfNonVLModel();
123
+ this.hasWarnedNonVLModel = true;
124
+ }
125
+ }
126
+ resolveReplanningCycleLimit(modelConfigForPlanning) {
127
+ if (void 0 !== this.opts.replanningCycleLimit) return this.opts.replanningCycleLimit;
128
+ return (0, util_js_namespaceObject.isUITars)(modelConfigForPlanning.modelFamily) ? defaultVlmUiTarsReplanningCycleLimit : (0, util_js_namespaceObject.isAutoGLM)(modelConfigForPlanning.modelFamily) ? defaultAutoGlmReplanningCycleLimit : defaultReplanningCycleLimit;
129
+ }
130
+ async getActionSpace() {
131
+ return this.fullActionSpace;
132
+ }
133
+ isRetryableContextError(_error) {
134
+ return false;
135
+ }
136
+ async getUIContext(action) {
137
+ this.ensureVLModelWarning();
138
+ if (this.frozenUIContext) {
139
+ debug('Using frozen page context for action:', action);
140
+ return this.frozenUIContext;
141
+ }
142
+ const { modelFamily } = this.modelConfigManager.getModelConfig('default');
143
+ const maxRetries = Agent.CONTEXT_RETRY_MAX;
144
+ for(let attempt = 0;; attempt++)try {
145
+ return await (0, external_utils_js_namespaceObject_1.commonContextParser)(this.interface, {
146
+ uploadServerUrl: this.modelConfigManager.getUploadTestServerUrl(),
147
+ screenshotShrinkFactor: this.opts.screenshotShrinkFactor,
148
+ modelFamily
149
+ });
150
+ } catch (error) {
151
+ if (attempt < maxRetries && this.isRetryableContextError(error)) {
152
+ debug(`retryable context error (attempt ${attempt + 1}/${maxRetries}), retrying in ${Agent.CONTEXT_RETRY_DELAY_MS}ms: ${error}`);
153
+ await new Promise((resolve)=>setTimeout(resolve, Agent.CONTEXT_RETRY_DELAY_MS));
154
+ continue;
155
+ }
156
+ throw error;
157
+ }
158
+ }
159
+ async _snapshotContext() {
160
+ return await this.getUIContext('locate');
161
+ }
162
+ async setAIActionContext(prompt) {
163
+ await this.setAIActContext(prompt);
164
+ }
165
+ async setAIActContext(prompt) {
166
+ if (this.aiActContext) console.warn('aiActContext is already set, and it is called again, will override the previous setting');
167
+ this.opts.aiActContext = prompt;
168
+ this.opts.aiActionContext = prompt;
169
+ }
170
+ resetDump() {
171
+ this.dump = new external_types_js_namespaceObject.GroupedActionDump({
172
+ sdkVersion: (0, external_utils_js_namespaceObject.getVersion)(),
173
+ groupName: this.opts.groupName,
174
+ groupDescription: this.opts.groupDescription,
175
+ executions: [],
176
+ modelBriefs: [],
177
+ deviceType: this.interface.interfaceType
178
+ });
179
+ this.executionDumpIndexByRunner = new WeakMap();
180
+ return this.dump;
181
+ }
182
+ appendExecutionDump(execution, runner) {
183
+ const currentDump = this.dump;
184
+ if (runner) {
185
+ const existingIndex = this.executionDumpIndexByRunner.get(runner);
186
+ if (void 0 !== existingIndex) {
187
+ currentDump.executions[existingIndex] = execution;
188
+ return;
189
+ }
190
+ currentDump.executions.push(execution);
191
+ this.executionDumpIndexByRunner.set(runner, currentDump.executions.length - 1);
192
+ return;
193
+ }
194
+ currentDump.executions.push(execution);
195
+ }
196
+ dumpDataString(opt) {
197
+ this.dump.groupName = this.opts.groupName;
198
+ this.dump.groupDescription = this.opts.groupDescription;
199
+ if (utils_namespaceObject.ifInBrowser || opt?.inlineScreenshots) return this.dump.serializeWithInlineScreenshots();
200
+ return this.dump.serialize();
201
+ }
202
+ reportHTMLString(opt) {
203
+ return (0, external_utils_js_namespaceObject.reportHTMLContent)(this.dumpDataString(opt));
204
+ }
205
+ writeOutActionDumps(executionDump) {
206
+ const exec = executionDump || this.lastExecutionDump;
207
+ if (exec) {
208
+ this.lastExecutionDump = exec;
209
+ this.reportGenerator.onExecutionUpdate(exec, this.getGroupMeta());
210
+ }
211
+ this.reportFile = this.reportGenerator.getReportPath();
212
+ }
213
+ getGroupMeta() {
214
+ return {
215
+ groupName: this.dump.groupName,
216
+ groupDescription: this.dump.groupDescription,
217
+ sdkVersion: this.dump.sdkVersion,
218
+ modelBriefs: this.dump.modelBriefs,
219
+ deviceType: this.dump.deviceType
220
+ };
221
+ }
222
+ async callbackOnTaskStartTip(task) {
223
+ const param = (0, external_ui_utils_js_namespaceObject.paramStr)(task);
224
+ const tip = param ? `${(0, external_ui_utils_js_namespaceObject.typeStr)(task)} - ${param}` : (0, external_ui_utils_js_namespaceObject.typeStr)(task);
225
+ if (this.onTaskStartTip) await this.onTaskStartTip(tip);
226
+ }
227
+ wrapActionInActionSpace(name) {
228
+ return async (param)=>await this.callActionInActionSpace(name, param);
229
+ }
230
+ async callActionInActionSpace(type, opt) {
231
+ debug('callActionInActionSpace', type, ',', opt);
232
+ const actionPlan = {
233
+ type: type,
234
+ param: opt || {},
235
+ thought: ''
236
+ };
237
+ debug('actionPlan', actionPlan);
238
+ const plans = [
239
+ actionPlan
240
+ ].filter(Boolean);
241
+ const title = (0, external_ui_utils_js_namespaceObject.taskTitleStr)(type, (0, external_ui_utils_js_namespaceObject.locateParamStr)(opt?.locate || {}));
242
+ const defaultIntentModelConfig = this.modelConfigManager.getModelConfig('default');
243
+ const modelConfigForPlanning = this.modelConfigManager.getModelConfig('planning');
244
+ const { output } = await this.taskExecutor.runPlans(title, plans, modelConfigForPlanning, defaultIntentModelConfig);
245
+ return output;
246
+ }
247
+ async aiTap(locatePrompt, opt) {
248
+ (0, utils_namespaceObject.assert)(locatePrompt, 'missing locate prompt for tap');
249
+ const detailedLocateParam = (0, external_yaml_index_js_namespaceObject.buildDetailedLocateParam)(locatePrompt, opt);
250
+ const fileChooserAccept = opt?.fileChooserAccept ? this.normalizeFileInput(opt.fileChooserAccept) : void 0;
251
+ return (0, external_tasks_js_namespaceObject.withFileChooser)(this.interface, fileChooserAccept, async ()=>this.callActionInActionSpace('Tap', {
252
+ locate: detailedLocateParam
253
+ }));
254
+ }
255
+ async aiRightClick(locatePrompt, opt) {
256
+ (0, utils_namespaceObject.assert)(locatePrompt, 'missing locate prompt for right click');
257
+ const detailedLocateParam = (0, external_yaml_index_js_namespaceObject.buildDetailedLocateParam)(locatePrompt, opt);
258
+ return this.callActionInActionSpace('RightClick', {
259
+ locate: detailedLocateParam
260
+ });
261
+ }
262
+ async aiDoubleClick(locatePrompt, opt) {
263
+ (0, utils_namespaceObject.assert)(locatePrompt, 'missing locate prompt for double click');
264
+ const detailedLocateParam = (0, external_yaml_index_js_namespaceObject.buildDetailedLocateParam)(locatePrompt, opt);
265
+ return this.callActionInActionSpace('DoubleClick', {
266
+ locate: detailedLocateParam
267
+ });
268
+ }
269
+ async aiHover(locatePrompt, opt) {
270
+ (0, utils_namespaceObject.assert)(locatePrompt, 'missing locate prompt for hover');
271
+ const detailedLocateParam = (0, external_yaml_index_js_namespaceObject.buildDetailedLocateParam)(locatePrompt, opt);
272
+ return this.callActionInActionSpace('Hover', {
273
+ locate: detailedLocateParam
274
+ });
275
+ }
276
+ async aiInput(locatePromptOrValue, locatePromptOrOpt, optOrUndefined) {
277
+ let value;
278
+ let locatePrompt;
279
+ let opt;
280
+ if ('object' == typeof locatePromptOrOpt && null !== locatePromptOrOpt && 'value' in locatePromptOrOpt) {
281
+ locatePrompt = locatePromptOrValue;
282
+ const optWithValue = locatePromptOrOpt;
283
+ value = optWithValue.value;
284
+ opt = optWithValue;
285
+ } else {
286
+ value = locatePromptOrValue;
287
+ locatePrompt = locatePromptOrOpt;
288
+ opt = {
289
+ ...optOrUndefined,
290
+ value
291
+ };
292
+ }
293
+ (0, utils_namespaceObject.assert)('string' == typeof value || 'number' == typeof value, 'input value must be a string or number, use empty string if you want to clear the input');
294
+ (0, utils_namespaceObject.assert)(locatePrompt, 'missing locate prompt for input');
295
+ const detailedLocateParam = (0, external_yaml_index_js_namespaceObject.buildDetailedLocateParam)(locatePrompt, opt);
296
+ const stringValue = 'number' == typeof value ? String(value) : value;
297
+ const mode = opt?.mode === 'append' ? 'typeOnly' : opt?.mode;
298
+ return this.callActionInActionSpace('Input', {
299
+ ...opt || {},
300
+ value: stringValue,
301
+ locate: detailedLocateParam,
302
+ mode
303
+ });
304
+ }
305
+ async aiKeyboardPress(locatePromptOrKeyName, locatePromptOrOpt, optOrUndefined) {
306
+ let keyName;
307
+ let locatePrompt;
308
+ let opt;
309
+ if ('object' == typeof locatePromptOrOpt && null !== locatePromptOrOpt && 'keyName' in locatePromptOrOpt) {
310
+ locatePrompt = locatePromptOrKeyName;
311
+ opt = locatePromptOrOpt;
312
+ } else {
313
+ keyName = locatePromptOrKeyName;
314
+ locatePrompt = locatePromptOrOpt;
315
+ opt = {
316
+ ...optOrUndefined || {},
317
+ keyName
318
+ };
319
+ }
320
+ (0, utils_namespaceObject.assert)(opt?.keyName, 'missing keyName for keyboard press');
321
+ const detailedLocateParam = locatePrompt ? (0, external_yaml_index_js_namespaceObject.buildDetailedLocateParam)(locatePrompt, opt) : void 0;
322
+ return this.callActionInActionSpace('KeyboardPress', {
323
+ ...opt || {},
324
+ locate: detailedLocateParam
325
+ });
326
+ }
327
+ async aiScroll(locatePromptOrScrollParam, locatePromptOrOpt, optOrUndefined) {
328
+ let scrollParam;
329
+ let locatePrompt;
330
+ let opt;
331
+ const isLocatePromptLike = (value)=>{
332
+ if ('string' == typeof value || void 0 === value) return true;
333
+ return 'object' == typeof value && null !== value && 'prompt' in value;
334
+ };
335
+ if (isLocatePromptLike(locatePromptOrScrollParam) && 'object' == typeof locatePromptOrOpt && null !== locatePromptOrOpt) {
336
+ locatePrompt = locatePromptOrScrollParam;
337
+ opt = locatePromptOrOpt;
338
+ } else {
339
+ scrollParam = locatePromptOrScrollParam;
340
+ locatePrompt = locatePromptOrOpt;
341
+ opt = {
342
+ ...optOrUndefined || {},
343
+ ...scrollParam || {}
344
+ };
345
+ }
346
+ if (opt) {
347
+ const normalizedScrollType = normalizeScrollType(opt.scrollType);
348
+ if (normalizedScrollType !== opt.scrollType) opt = {
349
+ ...opt || {},
350
+ scrollType: normalizedScrollType
351
+ };
352
+ }
353
+ const detailedLocateParam = (0, external_yaml_index_js_namespaceObject.buildDetailedLocateParam)(locatePrompt || '', opt);
354
+ return this.callActionInActionSpace('Scroll', {
355
+ ...opt || {},
356
+ locate: detailedLocateParam
357
+ });
358
+ }
359
+ async aiPinch(locatePrompt, opt) {
360
+ const detailedLocateParam = (0, external_yaml_index_js_namespaceObject.buildDetailedLocateParam)(locatePrompt || '', opt);
361
+ return this.callActionInActionSpace('Pinch', {
362
+ ...opt,
363
+ locate: detailedLocateParam
364
+ });
365
+ }
366
+ async aiAct(taskPrompt, opt) {
367
+ const fileChooserAccept = opt?.fileChooserAccept ? this.normalizeFileInput(opt.fileChooserAccept) : void 0;
368
+ const abortSignal = opt?.abortSignal;
369
+ if (abortSignal?.aborted) throw new Error(`aiAct aborted: ${abortSignal.reason || 'signal already aborted'}`);
370
+ const runAiAct = async ()=>{
371
+ const modelConfigForPlanning = this.modelConfigManager.getModelConfig('planning');
372
+ const defaultIntentModelConfig = this.modelConfigManager.getModelConfig('default');
373
+ const deepThink = opt?.deepThink === 'unset' ? void 0 : opt?.deepThink;
374
+ const deepLocate = opt?.deepLocate;
375
+ const noIndividualLocateModel = modelConfigForPlanning.modelName === defaultIntentModelConfig.modelName && modelConfigForPlanning.openaiBaseURL === defaultIntentModelConfig.openaiBaseURL;
376
+ const includeBboxInPlanning = !deepThink && noIndividualLocateModel;
377
+ debug('setting includeBboxInPlanning to', includeBboxInPlanning, {
378
+ deepThink,
379
+ noIndividualLocateModel
380
+ });
381
+ const cacheable = opt?.cacheable;
382
+ const replanningCycleLimit = this.resolveReplanningCycleLimit(modelConfigForPlanning);
383
+ const isVlmUiTars = (0, util_js_namespaceObject.isUITars)(modelConfigForPlanning.modelFamily);
384
+ const isAutoGlm = (0, util_js_namespaceObject.isAutoGLM)(modelConfigForPlanning.modelFamily);
385
+ const matchedCache = isVlmUiTars || isAutoGlm || false === cacheable ? void 0 : this.taskCache?.matchPlanCache(taskPrompt);
386
+ if (matchedCache?.cacheUsable && this.taskCache?.isCacheResultUsed && matchedCache.cacheContent?.yamlWorkflow?.trim()) {
387
+ await this.taskExecutor.loadYamlFlowAsPlanning(taskPrompt, matchedCache.cacheContent.yamlWorkflow);
388
+ debug('matched cache, will call .runYaml to run the action');
389
+ const yaml = matchedCache.cacheContent.yamlWorkflow;
390
+ await this.runYaml(yaml);
391
+ return;
392
+ }
393
+ const imagesIncludeCount = deepThink ? 2 : 1;
394
+ const { output: actionOutput } = await this.taskExecutor.action(taskPrompt, modelConfigForPlanning, defaultIntentModelConfig, includeBboxInPlanning, this.aiActContext, cacheable, replanningCycleLimit, imagesIncludeCount, deepThink, fileChooserAccept, deepLocate, abortSignal);
395
+ if (this.taskCache && actionOutput?.yamlFlow?.length && false !== cacheable) {
396
+ const yamlContent = {
397
+ tasks: [
398
+ {
399
+ name: taskPrompt,
400
+ flow: actionOutput.yamlFlow
401
+ }
402
+ ]
403
+ };
404
+ const yamlFlowStr = external_js_yaml_default().dump(yamlContent);
405
+ this.taskCache.updateOrAppendCacheRecord({
406
+ type: 'plan',
407
+ prompt: taskPrompt,
408
+ yamlWorkflow: yamlFlowStr
409
+ }, matchedCache);
410
+ }
411
+ return actionOutput?.output;
412
+ };
413
+ return await runAiAct();
414
+ }
415
+ async aiAction(taskPrompt, opt) {
416
+ return this.aiAct(taskPrompt, opt);
417
+ }
418
+ async aiQuery(demand, opt = defaultServiceExtractOption) {
419
+ const modelConfig = this.modelConfigManager.getModelConfig('insight');
420
+ const { output } = await this.taskExecutor.createTypeQueryExecution('Query', demand, modelConfig, opt);
421
+ return output;
422
+ }
423
+ async aiBoolean(prompt, opt = defaultServiceExtractOption) {
424
+ const modelConfig = this.modelConfigManager.getModelConfig('insight');
425
+ const { textPrompt, multimodalPrompt } = (0, external_utils_js_namespaceObject_1.parsePrompt)(prompt);
426
+ const { output } = await this.taskExecutor.createTypeQueryExecution('Boolean', textPrompt, modelConfig, opt, multimodalPrompt);
427
+ return output;
428
+ }
429
+ async aiNumber(prompt, opt = defaultServiceExtractOption) {
430
+ const modelConfig = this.modelConfigManager.getModelConfig('insight');
431
+ const { textPrompt, multimodalPrompt } = (0, external_utils_js_namespaceObject_1.parsePrompt)(prompt);
432
+ const { output } = await this.taskExecutor.createTypeQueryExecution('Number', textPrompt, modelConfig, opt, multimodalPrompt);
433
+ return output;
434
+ }
435
+ async aiString(prompt, opt = defaultServiceExtractOption) {
436
+ const modelConfig = this.modelConfigManager.getModelConfig('insight');
437
+ const { textPrompt, multimodalPrompt } = (0, external_utils_js_namespaceObject_1.parsePrompt)(prompt);
438
+ const { output } = await this.taskExecutor.createTypeQueryExecution('String', textPrompt, modelConfig, opt, multimodalPrompt);
439
+ return output;
440
+ }
441
+ async aiAsk(prompt, opt = defaultServiceExtractOption) {
442
+ return this.aiString(prompt, opt);
443
+ }
444
+ async describeElementAtPoint(center, opt) {
445
+ const { verifyPrompt = true, retryLimit = 3 } = opt || {};
446
+ let success = false;
447
+ let retryCount = 0;
448
+ let resultPrompt = '';
449
+ let deepLocate = opt?.deepLocate || false;
450
+ let verifyResult;
451
+ while(!success && retryCount < retryLimit){
452
+ if (retryCount >= 2) deepLocate = true;
453
+ debug('aiDescribe', center, 'verifyPrompt', verifyPrompt, 'retryCount', retryCount, 'deepLocate', deepLocate);
454
+ const modelConfig = this.modelConfigManager.getModelConfig('insight');
455
+ const text = await this.service.describe(center, modelConfig, {
456
+ deepLocate
457
+ });
458
+ debug('aiDescribe text', text);
459
+ (0, utils_namespaceObject.assert)(text.description, `failed to describe element at [${center}]`);
460
+ resultPrompt = text.description;
461
+ verifyResult = await this.verifyLocator(resultPrompt, void 0, center, opt);
462
+ if (verifyResult.pass) success = true;
463
+ else retryCount++;
464
+ }
465
+ return {
466
+ prompt: resultPrompt,
467
+ deepLocate,
468
+ verifyResult
469
+ };
470
+ }
471
+ async verifyLocator(prompt, locateOpt, expectCenter, verifyLocateOption) {
472
+ debug('verifyLocator', prompt, locateOpt, expectCenter, verifyLocateOption);
473
+ const { center: verifyCenter, rect: verifyRect } = await this.aiLocate(prompt, locateOpt);
474
+ const distance = distanceOfTwoPoints(expectCenter, verifyCenter);
475
+ const included = includedInRect(expectCenter, verifyRect);
476
+ const pass = distance <= (verifyLocateOption?.centerDistanceThreshold || 20) || included;
477
+ const verifyResult = {
478
+ pass,
479
+ rect: verifyRect,
480
+ center: verifyCenter,
481
+ centerDistance: distance
482
+ };
483
+ debug('aiDescribe verifyResult', verifyResult);
484
+ return verifyResult;
485
+ }
486
+ async aiLocate(prompt, opt) {
487
+ const locateParam = (0, external_yaml_index_js_namespaceObject.buildDetailedLocateParam)(prompt, opt);
488
+ (0, utils_namespaceObject.assert)(locateParam, 'cannot get locate param for aiLocate');
489
+ const locatePlan = (0, external_tasks_js_namespaceObject.locatePlanForLocate)(locateParam);
490
+ const plans = [
491
+ locatePlan
492
+ ];
493
+ const defaultIntentModelConfig = this.modelConfigManager.getModelConfig('default');
494
+ const modelConfigForPlanning = this.modelConfigManager.getModelConfig('planning');
495
+ const { output } = await this.taskExecutor.runPlans((0, external_ui_utils_js_namespaceObject.taskTitleStr)('Locate', (0, external_ui_utils_js_namespaceObject.locateParamStr)(locateParam)), plans, modelConfigForPlanning, defaultIntentModelConfig);
496
+ const { element } = output;
497
+ return {
498
+ rect: element?.rect,
499
+ center: element?.center,
500
+ dpr: element?.dpr
501
+ };
502
+ }
503
+ async aiAssert(assertion, msg, opt) {
504
+ const modelConfig = this.modelConfigManager.getModelConfig('insight');
505
+ const serviceOpt = {
506
+ domIncluded: opt?.domIncluded ?? defaultServiceExtractOption.domIncluded,
507
+ screenshotIncluded: opt?.screenshotIncluded ?? defaultServiceExtractOption.screenshotIncluded
508
+ };
509
+ const { textPrompt, multimodalPrompt } = (0, external_utils_js_namespaceObject_1.parsePrompt)(assertion);
510
+ const assertionText = 'string' == typeof assertion ? assertion : assertion.prompt;
511
+ try {
512
+ const { output, thought } = await this.taskExecutor.createTypeQueryExecution('Assert', textPrompt, modelConfig, serviceOpt, multimodalPrompt);
513
+ const pass = Boolean(output);
514
+ const message = pass ? void 0 : `Assertion failed: ${msg || assertionText}\nReason: ${thought || '(no_reason)'}`;
515
+ if (opt?.keepRawResponse) return {
516
+ pass,
517
+ thought,
518
+ message
519
+ };
520
+ if (!pass) throw new Error(message);
521
+ } catch (error) {
522
+ if (error instanceof external_tasks_js_namespaceObject.TaskExecutionError) {
523
+ const errorTask = error.errorTask;
524
+ const thought = errorTask?.thought;
525
+ const rawError = errorTask?.error;
526
+ const rawMessage = errorTask?.errorMessage || (rawError instanceof Error ? rawError.message : rawError ? String(rawError) : void 0);
527
+ const reason = thought || rawMessage || '(no_reason)';
528
+ const message = `Assertion failed: ${msg || assertionText}\nReason: ${reason}`;
529
+ if (opt?.keepRawResponse) return {
530
+ pass: false,
531
+ thought,
532
+ message
533
+ };
534
+ throw new Error(message, {
535
+ cause: rawError ?? error
536
+ });
537
+ }
538
+ throw error;
539
+ }
540
+ }
541
+ async aiWaitFor(assertion, opt) {
542
+ const modelConfig = this.modelConfigManager.getModelConfig('insight');
543
+ await this.taskExecutor.waitFor(assertion, {
544
+ ...opt,
545
+ timeoutMs: opt?.timeoutMs || 15000,
546
+ checkIntervalMs: opt?.checkIntervalMs || 3000
547
+ }, modelConfig);
548
+ }
549
+ async ai(...args) {
550
+ return this.aiAct(...args);
551
+ }
552
+ async runYaml(yamlScriptContent) {
553
+ const script = (0, external_yaml_index_js_namespaceObject.parseYamlScript)(yamlScriptContent, 'yaml');
554
+ const player = new external_yaml_index_js_namespaceObject.ScriptPlayer(script, async ()=>({
555
+ agent: this,
556
+ freeFn: []
557
+ }));
558
+ await player.run();
559
+ if ('error' === player.status) {
560
+ const errors = player.taskStatusList.filter((task)=>'error' === task.status).map((task)=>`task - ${task.name}: ${task.error?.message}`).join('\n');
561
+ throw new Error(`Error(s) occurred in running yaml script:\n${errors}`);
562
+ }
563
+ return {
564
+ result: player.result
565
+ };
566
+ }
567
+ async evaluateJavaScript(script) {
568
+ (0, utils_namespaceObject.assert)(this.interface.evaluateJavaScript, 'evaluateJavaScript is not supported in current agent');
569
+ return this.interface.evaluateJavaScript(script);
570
+ }
571
+ addDumpUpdateListener(listener) {
572
+ this.dumpUpdateListeners.push(listener);
573
+ return ()=>{
574
+ this.removeDumpUpdateListener(listener);
575
+ };
576
+ }
577
+ removeDumpUpdateListener(listener) {
578
+ const index = this.dumpUpdateListeners.indexOf(listener);
579
+ if (index > -1) this.dumpUpdateListeners.splice(index, 1);
580
+ }
581
+ clearDumpUpdateListeners() {
582
+ this.dumpUpdateListeners = [];
583
+ }
584
+ async destroy() {
585
+ if (this.destroyed) return;
586
+ await this.reportGenerator.flush();
587
+ const finalPath = await this.reportGenerator.finalize();
588
+ this.reportFile = finalPath;
589
+ await this.interface.destroy?.();
590
+ this.resetDump();
591
+ this.destroyed = true;
592
+ }
593
+ async recordToReport(title, opt) {
594
+ const base64 = await this.interface.screenshotBase64();
595
+ const now = Date.now();
596
+ const screenshot = external_screenshot_item_js_namespaceObject.ScreenshotItem.create(base64, now);
597
+ const recorder = [
598
+ {
599
+ type: 'screenshot',
600
+ ts: now,
601
+ screenshot
602
+ }
603
+ ];
604
+ const task = {
605
+ taskId: (0, utils_namespaceObject.uuid)(),
606
+ type: 'Log',
607
+ subType: 'Screenshot',
608
+ status: 'finished',
609
+ recorder,
610
+ timing: {
611
+ start: now,
612
+ end: now,
613
+ cost: 0
614
+ },
615
+ param: {
616
+ content: opt?.content || ''
617
+ },
618
+ executor: async ()=>{}
619
+ };
620
+ const executionDump = new external_types_js_namespaceObject.ExecutionDump({
621
+ id: (0, utils_namespaceObject.uuid)(),
622
+ logTime: now,
623
+ name: `Log - ${title || 'untitled'}`,
624
+ description: opt?.content || '',
625
+ tasks: [
626
+ task
627
+ ]
628
+ });
629
+ this.appendExecutionDump(executionDump);
630
+ const dumpString = this.dumpDataString();
631
+ for (const listener of this.dumpUpdateListeners)try {
632
+ listener(dumpString);
633
+ } catch (error) {
634
+ console.error('Error in onDumpUpdate listener', error);
635
+ }
636
+ this.writeOutActionDumps(executionDump);
637
+ await this.reportGenerator.flush();
638
+ }
639
+ async logScreenshot(title, opt) {
640
+ await this.recordToReport(title, opt);
641
+ }
642
+ _unstableLogContent() {
643
+ const { groupName, groupDescription, executions } = this.dump;
644
+ return {
645
+ groupName,
646
+ groupDescription,
647
+ executions: executions || []
648
+ };
649
+ }
650
+ async freezePageContext() {
651
+ debug('Freezing page context');
652
+ const context = await this._snapshotContext();
653
+ context._isFrozen = true;
654
+ this.frozenUIContext = context;
655
+ debug('Page context frozen successfully');
656
+ }
657
+ async unfreezePageContext() {
658
+ debug('Unfreezing page context');
659
+ this.frozenUIContext = void 0;
660
+ debug('Page context unfrozen successfully');
661
+ }
662
+ processCacheConfig(opts) {
663
+ if (true === opts.cache) throw new Error('cache: true requires an explicit cache ID. Please provide:\nExample: cache: { id: "my-cache-id" }');
664
+ if (opts.cache && 'object' == typeof opts.cache && null !== opts.cache && !opts.cache.id) throw new Error('cache configuration requires an explicit id.\nExample: cache: { id: "my-cache-id" }');
665
+ const cacheConfig = (0, external_utils_js_namespaceObject.processCacheConfig)(opts.cache, opts.cacheId || opts.testId || 'default');
666
+ if (!cacheConfig) return null;
667
+ if ('object' == typeof cacheConfig && null !== cacheConfig) {
668
+ const id = cacheConfig.id;
669
+ const rawStrategy = cacheConfig.strategy;
670
+ let strategyValue;
671
+ if (void 0 === rawStrategy) strategyValue = 'read-write';
672
+ else if ('string' == typeof rawStrategy) strategyValue = rawStrategy;
673
+ else throw new Error(`cache.strategy must be a string when provided, but received type ${typeof rawStrategy}`);
674
+ if (!isValidCacheStrategy(strategyValue)) throw new Error(`cache.strategy must be one of ${CACHE_STRATEGY_VALUES}, but received "${strategyValue}"`);
675
+ const isReadOnly = 'read-only' === strategyValue;
676
+ const isWriteOnly = 'write-only' === strategyValue;
677
+ return {
678
+ id,
679
+ enabled: !isWriteOnly,
680
+ readOnly: isReadOnly,
681
+ writeOnly: isWriteOnly
682
+ };
683
+ }
684
+ return null;
685
+ }
686
+ normalizeFilePaths(files) {
687
+ if (utils_namespaceObject.ifInBrowser) throw new Error('File chooser is not supported in browser environment');
688
+ return files.map((file)=>{
689
+ const absolutePath = (0, external_node_path_namespaceObject.resolve)(file);
690
+ if (!(0, external_node_fs_namespaceObject.existsSync)(absolutePath)) throw new Error(`File not found: ${file}`);
691
+ return absolutePath;
692
+ });
693
+ }
694
+ normalizeFileInput(files) {
695
+ const filesArray = Array.isArray(files) ? files : [
696
+ files
697
+ ];
698
+ return this.normalizeFilePaths(filesArray);
699
+ }
700
+ async flushCache(options) {
701
+ if (!this.taskCache) throw new Error('Cache is not configured');
702
+ this.taskCache.flushCacheToFile(options);
703
+ }
704
+ constructor(interfaceInstance, opts){
705
+ _define_property(this, "interface", void 0);
706
+ _define_property(this, "service", void 0);
707
+ _define_property(this, "dump", void 0);
708
+ _define_property(this, "reportFile", void 0);
709
+ _define_property(this, "reportFileName", void 0);
710
+ _define_property(this, "taskExecutor", void 0);
711
+ _define_property(this, "opts", void 0);
712
+ _define_property(this, "dryMode", false);
713
+ _define_property(this, "onTaskStartTip", void 0);
714
+ _define_property(this, "taskCache", void 0);
715
+ _define_property(this, "dumpUpdateListeners", []);
716
+ _define_property(this, "destroyed", false);
717
+ _define_property(this, "modelConfigManager", void 0);
718
+ _define_property(this, "frozenUIContext", void 0);
719
+ _define_property(this, "hasWarnedNonVLModel", false);
720
+ _define_property(this, "executionDumpIndexByRunner", new WeakMap());
721
+ _define_property(this, "fullActionSpace", void 0);
722
+ _define_property(this, "reportGenerator", void 0);
723
+ _define_property(this, "lastExecutionDump", void 0);
724
+ this.interface = interfaceInstance;
725
+ const envReplanningCycleLimit = env_namespaceObject.globalConfigManager.getEnvConfigValueAsNumber(env_namespaceObject.MIDSCENE_REPLANNING_CYCLE_LIMIT);
726
+ this.opts = Object.assign({
727
+ generateReport: true,
728
+ autoPrintReportMsg: true,
729
+ groupName: 'Midscene Report',
730
+ groupDescription: ''
731
+ }, opts || {}, opts?.replanningCycleLimit !== void 0 || void 0 === envReplanningCycleLimit || Number.isNaN(envReplanningCycleLimit) ? {} : {
732
+ replanningCycleLimit: envReplanningCycleLimit
733
+ });
734
+ const resolvedAiActContext = this.opts.aiActContext ?? this.opts.aiActionContext;
735
+ if (void 0 !== resolvedAiActContext) {
736
+ this.opts.aiActContext = resolvedAiActContext;
737
+ this.opts.aiActionContext ??= resolvedAiActContext;
738
+ }
739
+ if (opts?.modelConfig && ('object' != typeof opts?.modelConfig || Array.isArray(opts.modelConfig))) throw new Error(`opts.modelConfig must be a plain object map of env keys to values, but got ${typeof opts?.modelConfig}`);
740
+ const hasCustomConfig = opts?.modelConfig || opts?.createOpenAIClient;
741
+ this.modelConfigManager = hasCustomConfig ? new env_namespaceObject.ModelConfigManager(opts?.modelConfig, opts?.createOpenAIClient) : env_namespaceObject.globalModelConfigManager;
742
+ this.onTaskStartTip = this.opts.onTaskStartTip;
743
+ this.service = new (index_js_default())(async ()=>this.getUIContext());
744
+ const cacheConfigObj = this.processCacheConfig(opts || {});
745
+ if (cacheConfigObj) this.taskCache = new external_task_cache_js_namespaceObject.TaskCache(cacheConfigObj.id, cacheConfigObj.enabled, void 0, {
746
+ readOnly: cacheConfigObj.readOnly,
747
+ writeOnly: cacheConfigObj.writeOnly
748
+ });
749
+ const baseActionSpace = this.interface.actionSpace();
750
+ this.fullActionSpace = [
751
+ ...baseActionSpace,
752
+ (0, external_device_index_js_namespaceObject.defineActionSleep)()
753
+ ];
754
+ this.taskExecutor = new external_tasks_js_namespaceObject.TaskExecutor(this.interface, this.service, {
755
+ taskCache: this.taskCache,
756
+ onTaskStart: this.callbackOnTaskStartTip.bind(this),
757
+ replanningCycleLimit: this.opts.replanningCycleLimit,
758
+ waitAfterAction: this.opts.waitAfterAction,
759
+ useDeviceTimestamp: this.opts.useDeviceTimestamp,
760
+ actionSpace: this.fullActionSpace,
761
+ hooks: {
762
+ onTaskUpdate: (runner)=>{
763
+ const executionDump = runner.dump();
764
+ this.appendExecutionDump(executionDump, runner);
765
+ const dumpString = this.dumpDataString();
766
+ for (const listener of this.dumpUpdateListeners)try {
767
+ listener(dumpString, executionDump);
768
+ } catch (error) {
769
+ console.error('Error in onDumpUpdate listener', error);
770
+ }
771
+ this.writeOutActionDumps(executionDump);
772
+ }
773
+ }
774
+ });
775
+ this.dump = this.resetDump();
776
+ this.reportFileName = opts?.reportFileName || (0, external_utils_js_namespaceObject_1.getReportFileName)(opts?.testId || this.interface.interfaceType || 'web');
777
+ this.reportGenerator = external_report_generator_js_namespaceObject.ReportGenerator.create(this.reportFileName, {
778
+ generateReport: this.opts.generateReport,
779
+ outputFormat: this.opts.outputFormat,
780
+ autoPrintReportMsg: this.opts.autoPrintReportMsg
781
+ });
782
+ }
783
+ }
784
+ _define_property(Agent, "CONTEXT_RETRY_MAX", 3);
785
+ _define_property(Agent, "CONTEXT_RETRY_DELAY_MS", 1500);
786
+ const createAgent = (interfaceInstance, opts)=>new Agent(interfaceInstance, opts);
787
+ exports.Agent = __webpack_exports__.Agent;
788
+ exports.createAgent = __webpack_exports__.createAgent;
789
+ for(var __rspack_i in __webpack_exports__)if (-1 === [
790
+ "Agent",
791
+ "createAgent"
792
+ ].indexOf(__rspack_i)) exports[__rspack_i] = __webpack_exports__[__rspack_i];
793
+ Object.defineProperty(exports, '__esModule', {
794
+ value: true
795
+ });
796
+
797
+ //# sourceMappingURL=agent.js.map