@aiscene/core 1.6.2

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 +424 -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 +169 -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 +467 -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 +209 -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 +215 -0
  279. package/package.json +111 -0
@@ -0,0 +1,198 @@
1
+ import { ScreenshotItem } from "../screenshot-item.mjs";
2
+ import { uploadTestInfoToServer } from "../utils.mjs";
3
+ import { MIDSCENE_REPORT_QUIET, MIDSCENE_REPORT_TAG_NAME, globalConfigManager } from "@midscene/shared/env";
4
+ import { generateElementByRect } from "@midscene/shared/extractor";
5
+ import { imageInfoOfBase64, resizeImgBase64 } from "@midscene/shared/img";
6
+ import { getDebug } from "@midscene/shared/logger";
7
+ import { assert, logMsg, uuid } from "@midscene/shared/utils";
8
+ import dayjs from "dayjs";
9
+ import { debug as external_task_cache_mjs_debug } from "./task-cache.mjs";
10
+ async function commonContextParser(interfaceInstance, _opt) {
11
+ const debug = getDebug('commonContextParser');
12
+ assert(interfaceInstance, 'interfaceInstance is required');
13
+ debug("Getting interface description");
14
+ const description = interfaceInstance.describe?.() || '';
15
+ debug("Interface description end");
16
+ debug('Uploading test info to server');
17
+ uploadTestInfoToServer({
18
+ testUrl: description,
19
+ serverUrl: _opt.uploadServerUrl
20
+ });
21
+ debug('UploadTestInfoToServer end');
22
+ debug('will get size');
23
+ const interfaceSize = await interfaceInstance.size();
24
+ const { width: logicalWidth, height: logicalHeight } = interfaceSize;
25
+ if (interfaceSize.dpr) console.warn('Warning: return value of interface.size() include a dpr property, which is not expected and ignored. ');
26
+ if (!Number.isFinite(logicalWidth) || !Number.isFinite(logicalHeight)) throw new Error(`Invalid interface size: width and height must be finite numbers. Received width: ${logicalWidth}, height: ${logicalHeight}`);
27
+ if (logicalWidth <= 0 || logicalHeight <= 0) throw new Error(`Invalid interface size: width and height must be positive numbers. Received width: ${logicalWidth}, height: ${logicalHeight}`);
28
+ debug(`size: ${logicalWidth}x${logicalHeight}`);
29
+ const screenshotBase64 = await interfaceInstance.screenshotBase64();
30
+ const screenshotCapturedAt = Date.now();
31
+ assert(screenshotBase64, 'screenshotBase64 is required');
32
+ debug('will get screenshot dimensions');
33
+ const { width: imgWidth, height: imgHeight } = await imageInfoOfBase64(screenshotBase64);
34
+ if (!Number.isFinite(imgWidth) || !Number.isFinite(imgHeight)) throw new Error(`Invalid screenshot dimensions: width and height must be finite numbers. Received width: ${imgWidth}, height: ${imgHeight}`);
35
+ if (imgWidth <= 0 || imgHeight <= 0) throw new Error(`Invalid screenshot dimensions: width and height must be positive numbers. Received width: ${imgWidth}, height: ${imgHeight}`);
36
+ debug('screenshot dimensions', imgWidth, 'x', imgHeight);
37
+ const logicalIsPortrait = logicalWidth < logicalHeight;
38
+ const screenshotIsPortrait = imgWidth < imgHeight;
39
+ let finalLogicalWidth = logicalWidth;
40
+ if (logicalIsPortrait !== screenshotIsPortrait) {
41
+ debug(`Orientation mismatch detected: logical size ${logicalWidth}x${logicalHeight} (${logicalIsPortrait ? 'portrait' : 'landscape'}) vs screenshot ${imgWidth}x${imgHeight} (${screenshotIsPortrait ? 'portrait' : 'landscape'}). Swapping logical dimensions.`);
42
+ finalLogicalWidth = logicalHeight;
43
+ }
44
+ const userShrinkFactor = _opt.screenshotShrinkFactor ?? 1;
45
+ if (!Number.isFinite(userShrinkFactor) || userShrinkFactor < 1) throw new Error(`Invalid screenshotShrinkFactor: must be a finite number >= 1. Received: ${userShrinkFactor}`);
46
+ const dpr = imgWidth / finalLogicalWidth;
47
+ debug('calculated dpr:', dpr);
48
+ const shrunkShotToLogicalRatio = dpr / userShrinkFactor;
49
+ debug('shrunkShotToLogicalRatio', shrunkShotToLogicalRatio);
50
+ if (1 !== userShrinkFactor) {
51
+ const targetWidth = Math.round(imgWidth / userShrinkFactor);
52
+ const targetHeight = Math.round(imgHeight / userShrinkFactor);
53
+ debug(`Applying screenshot shrink factor: ${userShrinkFactor} (physical: ${imgWidth}x${imgHeight} -> target: ${targetWidth}x${targetHeight})`);
54
+ const resizedBase64 = await resizeImgBase64(screenshotBase64, {
55
+ width: targetWidth,
56
+ height: targetHeight
57
+ });
58
+ return {
59
+ shotSize: {
60
+ width: targetWidth,
61
+ height: targetHeight
62
+ },
63
+ deprecatedDpr: dpr,
64
+ screenshot: ScreenshotItem.create(resizedBase64, screenshotCapturedAt),
65
+ shrunkShotToLogicalRatio
66
+ };
67
+ }
68
+ return {
69
+ shotSize: {
70
+ width: imgWidth,
71
+ height: imgHeight
72
+ },
73
+ deprecatedDpr: dpr,
74
+ screenshot: ScreenshotItem.create(screenshotBase64, screenshotCapturedAt),
75
+ shrunkShotToLogicalRatio
76
+ };
77
+ }
78
+ function getReportFileName(tag = 'web') {
79
+ const reportTagName = globalConfigManager.getEnvConfigValue(MIDSCENE_REPORT_TAG_NAME);
80
+ const dateTimeInFileName = dayjs().format('YYYY-MM-DD_HH-mm-ss');
81
+ const uniqueId = uuid().substring(0, 8);
82
+ return `${reportTagName || tag}-${dateTimeInFileName}-${uniqueId}`;
83
+ }
84
+ function printReportMsg(filepath) {
85
+ if (globalConfigManager.getEnvConfigInBoolean(MIDSCENE_REPORT_QUIET)) return;
86
+ logMsg(`Midscene - report file updated: ${filepath}`);
87
+ }
88
+ function getCurrentExecutionFile(trace) {
89
+ const error = new Error();
90
+ const stackTrace = trace || error.stack;
91
+ const pkgDir = process.cwd() || '';
92
+ if (stackTrace) {
93
+ const stackLines = stackTrace.split('\n');
94
+ for (const line of stackLines)if (line.includes('.spec.') || line.includes('.test.') || line.includes('.ts') || line.includes('.js')) {
95
+ const match = line.match(/(?:at\s+)?(.*?\.(?:spec|test)\.[jt]s)/);
96
+ if (match?.[1]) {
97
+ const targetFileName = match[1].replace(pkgDir, '').trim().replace('at ', '');
98
+ return targetFileName;
99
+ }
100
+ }
101
+ }
102
+ return false;
103
+ }
104
+ const testFileIndex = new Map();
105
+ function generateCacheId(fileName) {
106
+ let taskFile = fileName || getCurrentExecutionFile();
107
+ if (!taskFile) {
108
+ taskFile = uuid();
109
+ console.warn('Midscene - using random UUID for cache id. Cache may be invalid.');
110
+ }
111
+ if (testFileIndex.has(taskFile)) {
112
+ const currentIndex = testFileIndex.get(taskFile);
113
+ if (void 0 !== currentIndex) testFileIndex.set(taskFile, currentIndex + 1);
114
+ } else testFileIndex.set(taskFile, 1);
115
+ return `${taskFile}-${testFileIndex.get(taskFile)}`;
116
+ }
117
+ function ifPlanLocateParamIsBbox(planLocateParam) {
118
+ return !!(planLocateParam.bbox && Array.isArray(planLocateParam.bbox) && 4 === planLocateParam.bbox.length);
119
+ }
120
+ function matchElementFromPlan(planLocateParam) {
121
+ if (!planLocateParam) return;
122
+ if (planLocateParam.bbox) {
123
+ const rect = {
124
+ left: planLocateParam.bbox[0],
125
+ top: planLocateParam.bbox[1],
126
+ width: planLocateParam.bbox[2] - planLocateParam.bbox[0] + 1,
127
+ height: planLocateParam.bbox[3] - planLocateParam.bbox[1] + 1
128
+ };
129
+ const element = generateElementByRect(rect, 'string' == typeof planLocateParam.prompt ? planLocateParam.prompt : planLocateParam.prompt?.prompt || '');
130
+ return element;
131
+ }
132
+ }
133
+ async function matchElementFromCache(context, cacheEntry, cachePrompt, cacheable) {
134
+ if (!cacheEntry) return;
135
+ if (false === cacheable) return void external_task_cache_mjs_debug('cache disabled for prompt: %s', cachePrompt);
136
+ if (!context.taskCache?.isCacheResultUsed) return;
137
+ if (!context.interfaceInstance.rectMatchesCacheFeature) return void external_task_cache_mjs_debug('interface does not implement rectMatchesCacheFeature, skip cache');
138
+ try {
139
+ const rect = await context.interfaceInstance.rectMatchesCacheFeature(cacheEntry);
140
+ const element = {
141
+ center: [
142
+ Math.round(rect.left + rect.width / 2),
143
+ Math.round(rect.top + rect.height / 2)
144
+ ],
145
+ rect,
146
+ description: 'string' == typeof cachePrompt ? cachePrompt : cachePrompt.prompt || ''
147
+ };
148
+ external_task_cache_mjs_debug('cache hit, prompt: %s', cachePrompt);
149
+ return element;
150
+ } catch (error) {
151
+ external_task_cache_mjs_debug('rectMatchesCacheFeature error: %s', error);
152
+ return;
153
+ }
154
+ }
155
+ const getMidsceneVersion = ()=>"1.6.2";
156
+ const parsePrompt = (prompt)=>{
157
+ if ('string' == typeof prompt) return {
158
+ textPrompt: prompt,
159
+ multimodalPrompt: void 0
160
+ };
161
+ return {
162
+ textPrompt: prompt.prompt,
163
+ multimodalPrompt: prompt.images ? {
164
+ images: prompt.images,
165
+ convertHttpImage2Base64: !!prompt.convertHttpImage2Base64
166
+ } : void 0
167
+ };
168
+ };
169
+ const transformLogicalElementToScreenshot = (element, shrunkShotToLogicalRatio)=>{
170
+ if (1 === shrunkShotToLogicalRatio) return element;
171
+ return {
172
+ ...element,
173
+ center: [
174
+ Math.round(element.center[0] * shrunkShotToLogicalRatio),
175
+ Math.round(element.center[1] * shrunkShotToLogicalRatio)
176
+ ],
177
+ rect: {
178
+ ...element.rect,
179
+ left: Math.round(element.rect.left * shrunkShotToLogicalRatio),
180
+ top: Math.round(element.rect.top * shrunkShotToLogicalRatio),
181
+ width: Math.round(element.rect.width * shrunkShotToLogicalRatio),
182
+ height: Math.round(element.rect.height * shrunkShotToLogicalRatio)
183
+ }
184
+ };
185
+ };
186
+ const transformLogicalRectToScreenshotRect = (rect, shrunkShotToLogicalRatio)=>{
187
+ if (1 === shrunkShotToLogicalRatio) return rect;
188
+ return {
189
+ ...rect,
190
+ left: Math.round(rect.left * shrunkShotToLogicalRatio),
191
+ top: Math.round(rect.top * shrunkShotToLogicalRatio),
192
+ width: Math.round(rect.width * shrunkShotToLogicalRatio),
193
+ height: Math.round(rect.height * shrunkShotToLogicalRatio)
194
+ };
195
+ };
196
+ export { commonContextParser, generateCacheId, getCurrentExecutionFile, getMidsceneVersion, getReportFileName, ifPlanLocateParamIsBbox, matchElementFromCache, matchElementFromPlan, parsePrompt, printReportMsg, transformLogicalElementToScreenshot, transformLogicalRectToScreenshotRect };
197
+
198
+ //# sourceMappingURL=utils.mjs.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"agent/utils.mjs","sources":["../../../src/agent/utils.ts"],"sourcesContent":["import type { TMultimodalPrompt, TUserPrompt } from '@/common';\nimport type { AbstractInterface } from '@/device';\nimport { ScreenshotItem } from '@/screenshot-item';\nimport type {\n ElementCacheFeature,\n LocateResultElement,\n PlanningLocateParam,\n Rect,\n UIContext,\n} from '@/types';\nimport { uploadTestInfoToServer } from '@/utils';\nimport type { TModelFamily } from '@midscene/shared/env';\nimport {\n MIDSCENE_REPORT_QUIET,\n MIDSCENE_REPORT_TAG_NAME,\n globalConfigManager,\n} from '@midscene/shared/env';\nimport { generateElementByRect } from '@midscene/shared/extractor';\nimport { imageInfoOfBase64, resizeImgBase64 } from '@midscene/shared/img';\nimport { getDebug } from '@midscene/shared/logger';\nimport { _keyDefinitions } from '@midscene/shared/us-keyboard-layout';\nimport { assert, logMsg, uuid } from '@midscene/shared/utils';\nimport dayjs from 'dayjs';\nimport type { TaskCache } from './task-cache';\nimport { debug as cacheDebug } from './task-cache';\n\nexport async function commonContextParser(\n interfaceInstance: AbstractInterface,\n _opt: {\n uploadServerUrl?: string;\n screenshotShrinkFactor?: number;\n modelFamily?: TModelFamily;\n },\n): Promise<UIContext> {\n const debug = getDebug('commonContextParser');\n\n assert(interfaceInstance, 'interfaceInstance is required');\n\n debug('Getting interface description');\n const description = interfaceInstance.describe?.() || '';\n debug('Interface description end');\n\n debug('Uploading test info to server');\n uploadTestInfoToServer({\n testUrl: description,\n serverUrl: _opt.uploadServerUrl,\n });\n debug('UploadTestInfoToServer end');\n\n debug('will get size');\n const interfaceSize = await interfaceInstance.size();\n const { width: logicalWidth, height: logicalHeight } = interfaceSize;\n\n if ((interfaceSize as unknown as { dpr: number }).dpr) {\n console.warn(\n 'Warning: return value of interface.size() include a dpr property, which is not expected and ignored. ',\n );\n }\n\n if (!Number.isFinite(logicalWidth) || !Number.isFinite(logicalHeight)) {\n throw new Error(\n `Invalid interface size: width and height must be finite numbers. Received width: ${logicalWidth}, height: ${logicalHeight}`,\n );\n }\n\n if (logicalWidth <= 0 || logicalHeight <= 0) {\n throw new Error(\n `Invalid interface size: width and height must be positive numbers. Received width: ${logicalWidth}, height: ${logicalHeight}`,\n );\n }\n\n debug(`size: ${logicalWidth}x${logicalHeight}`);\n\n const screenshotBase64 = await interfaceInstance.screenshotBase64();\n const screenshotCapturedAt = Date.now();\n assert(screenshotBase64!, 'screenshotBase64 is required');\n\n // Get physical screenshot dimensions\n debug('will get screenshot dimensions');\n const { width: imgWidth, height: imgHeight } =\n await imageInfoOfBase64(screenshotBase64);\n\n if (!Number.isFinite(imgWidth) || !Number.isFinite(imgHeight)) {\n throw new Error(\n `Invalid screenshot dimensions: width and height must be finite numbers. Received width: ${imgWidth}, height: ${imgHeight}`,\n );\n }\n if (imgWidth <= 0 || imgHeight <= 0) {\n throw new Error(\n `Invalid screenshot dimensions: width and height must be positive numbers. Received width: ${imgWidth}, height: ${imgHeight}`,\n );\n }\n debug('screenshot dimensions', imgWidth, 'x', imgHeight);\n\n // Detect orientation mismatch between logical size and screenshot.\n // Some devices (e.g. OPPO) report wrong orientation via ADB, causing\n // size() to return portrait dimensions even when the device is landscape.\n // We detect this by comparing aspect ratios and swap if they disagree.\n const logicalIsPortrait = logicalWidth < logicalHeight;\n const screenshotIsPortrait = imgWidth < imgHeight;\n let finalLogicalWidth = logicalWidth;\n let finalLogicalHeight = logicalHeight;\n if (logicalIsPortrait !== screenshotIsPortrait) {\n debug(\n `Orientation mismatch detected: logical size ${logicalWidth}x${logicalHeight} (${logicalIsPortrait ? 'portrait' : 'landscape'}) vs screenshot ${imgWidth}x${imgHeight} (${screenshotIsPortrait ? 'portrait' : 'landscape'}). Swapping logical dimensions.`,\n );\n finalLogicalWidth = logicalHeight;\n finalLogicalHeight = logicalWidth;\n }\n\n const userShrinkFactor = _opt.screenshotShrinkFactor ?? 1;\n\n if (!Number.isFinite(userShrinkFactor) || userShrinkFactor < 1) {\n throw new Error(\n `Invalid screenshotShrinkFactor: must be a finite number >= 1. Received: ${userShrinkFactor}`,\n );\n }\n\n const dpr = imgWidth / finalLogicalWidth;\n\n debug('calculated dpr:', dpr);\n\n const shrunkShotToLogicalRatio = dpr / userShrinkFactor;\n\n debug('shrunkShotToLogicalRatio', shrunkShotToLogicalRatio);\n\n if (userShrinkFactor !== 1) {\n const targetWidth = Math.round(imgWidth / userShrinkFactor);\n const targetHeight = Math.round(imgHeight / userShrinkFactor);\n\n debug(\n `Applying screenshot shrink factor: ${userShrinkFactor} (physical: ${imgWidth}x${imgHeight} -> target: ${targetWidth}x${targetHeight})`,\n );\n\n const resizedBase64 = await resizeImgBase64(screenshotBase64, {\n width: targetWidth,\n height: targetHeight,\n });\n return {\n shotSize: {\n width: targetWidth,\n height: targetHeight,\n },\n deprecatedDpr: dpr,\n screenshot: ScreenshotItem.create(resizedBase64, screenshotCapturedAt),\n shrunkShotToLogicalRatio,\n };\n }\n\n return {\n shotSize: {\n width: imgWidth,\n height: imgHeight,\n },\n deprecatedDpr: dpr,\n screenshot: ScreenshotItem.create(screenshotBase64, screenshotCapturedAt),\n shrunkShotToLogicalRatio,\n };\n}\n\nexport function getReportFileName(tag = 'web') {\n const reportTagName = globalConfigManager.getEnvConfigValue(\n MIDSCENE_REPORT_TAG_NAME,\n );\n const dateTimeInFileName = dayjs().format('YYYY-MM-DD_HH-mm-ss');\n // ensure uniqueness at the same time\n const uniqueId = uuid().substring(0, 8);\n return `${reportTagName || tag}-${dateTimeInFileName}-${uniqueId}`;\n}\n\nexport function printReportMsg(filepath: string) {\n if (globalConfigManager.getEnvConfigInBoolean(MIDSCENE_REPORT_QUIET)) {\n return;\n }\n logMsg(`Midscene - report file updated: ${filepath}`);\n}\n\n/**\n * Get the current execution file name\n * @returns The name of the current execution file\n */\nexport function getCurrentExecutionFile(trace?: string): string | false {\n const error = new Error();\n const stackTrace = trace || error.stack;\n const pkgDir = process.cwd() || '';\n if (stackTrace) {\n const stackLines = stackTrace.split('\\n');\n for (const line of stackLines) {\n if (\n line.includes('.spec.') ||\n line.includes('.test.') ||\n line.includes('.ts') ||\n line.includes('.js')\n ) {\n const match = line.match(/(?:at\\s+)?(.*?\\.(?:spec|test)\\.[jt]s)/);\n if (match?.[1]) {\n const targetFileName = match[1]\n .replace(pkgDir, '')\n .trim()\n .replace('at ', '');\n return targetFileName;\n }\n }\n }\n }\n return false;\n}\n\nconst testFileIndex = new Map<string, number>();\n\nexport function generateCacheId(fileName?: string): string {\n let taskFile = fileName || getCurrentExecutionFile();\n if (!taskFile) {\n taskFile = uuid();\n console.warn(\n 'Midscene - using random UUID for cache id. Cache may be invalid.',\n );\n }\n\n if (testFileIndex.has(taskFile)) {\n const currentIndex = testFileIndex.get(taskFile);\n if (currentIndex !== undefined) {\n testFileIndex.set(taskFile, currentIndex + 1);\n }\n } else {\n testFileIndex.set(taskFile, 1);\n }\n return `${taskFile}-${testFileIndex.get(taskFile)}`;\n}\n\nexport function ifPlanLocateParamIsBbox(\n planLocateParam: PlanningLocateParam,\n): boolean {\n return !!(\n planLocateParam.bbox &&\n Array.isArray(planLocateParam.bbox) &&\n planLocateParam.bbox.length === 4\n );\n}\n\nexport function matchElementFromPlan(\n planLocateParam: PlanningLocateParam,\n): LocateResultElement | undefined {\n if (!planLocateParam) {\n return undefined;\n }\n\n if (planLocateParam.bbox) {\n // Convert bbox [x1, y1, x2, y2] to rect {left, top, width, height}\n const rect = {\n left: planLocateParam.bbox[0],\n top: planLocateParam.bbox[1],\n width: planLocateParam.bbox[2] - planLocateParam.bbox[0] + 1,\n height: planLocateParam.bbox[3] - planLocateParam.bbox[1] + 1,\n };\n\n const element = generateElementByRect(\n rect,\n typeof planLocateParam.prompt === 'string'\n ? planLocateParam.prompt\n : planLocateParam.prompt?.prompt || '',\n );\n return element;\n }\n\n return undefined;\n}\n\nexport async function matchElementFromCache(\n context: {\n taskCache?: TaskCache;\n interfaceInstance: AbstractInterface;\n },\n cacheEntry: ElementCacheFeature | undefined,\n cachePrompt: TUserPrompt,\n cacheable: boolean | undefined,\n): Promise<LocateResultElement | undefined> {\n if (!cacheEntry) {\n return undefined;\n }\n\n if (cacheable === false) {\n cacheDebug('cache disabled for prompt: %s', cachePrompt);\n return undefined;\n }\n\n if (!context.taskCache?.isCacheResultUsed) {\n return undefined;\n }\n\n if (!context.interfaceInstance.rectMatchesCacheFeature) {\n cacheDebug(\n 'interface does not implement rectMatchesCacheFeature, skip cache',\n );\n return undefined;\n }\n\n try {\n const rect =\n await context.interfaceInstance.rectMatchesCacheFeature(cacheEntry);\n const element: LocateResultElement = {\n center: [\n Math.round(rect.left + rect.width / 2),\n Math.round(rect.top + rect.height / 2),\n ],\n rect,\n description:\n typeof cachePrompt === 'string'\n ? cachePrompt\n : cachePrompt.prompt || '',\n };\n\n cacheDebug('cache hit, prompt: %s', cachePrompt);\n return element;\n } catch (error) {\n cacheDebug('rectMatchesCacheFeature error: %s', error);\n return undefined;\n }\n}\n\ndeclare const __VERSION__: string | undefined;\n\nexport const getMidsceneVersion = (): string => {\n if (typeof __VERSION__ !== 'undefined') {\n return __VERSION__;\n } else if (\n process.env.__VERSION__ &&\n process.env.__VERSION__ !== 'undefined'\n ) {\n return process.env.__VERSION__;\n }\n throw new Error('__VERSION__ inject failed during build');\n};\n\nexport const parsePrompt = (\n prompt: TUserPrompt,\n): {\n textPrompt: string;\n multimodalPrompt?: TMultimodalPrompt;\n} => {\n if (typeof prompt === 'string') {\n return {\n textPrompt: prompt,\n multimodalPrompt: undefined,\n };\n }\n return {\n textPrompt: prompt.prompt,\n multimodalPrompt: prompt.images\n ? {\n images: prompt.images,\n convertHttpImage2Base64: !!prompt.convertHttpImage2Base64,\n }\n : undefined,\n };\n};\n\nexport const transformLogicalElementToScreenshot = (\n element: LocateResultElement,\n shrunkShotToLogicalRatio: number,\n): LocateResultElement => {\n if (shrunkShotToLogicalRatio === 1) {\n return element;\n }\n\n return {\n ...element,\n center: [\n Math.round(element.center[0] * shrunkShotToLogicalRatio),\n Math.round(element.center[1] * shrunkShotToLogicalRatio),\n ],\n rect: {\n ...element.rect,\n left: Math.round(element.rect.left * shrunkShotToLogicalRatio),\n top: Math.round(element.rect.top * shrunkShotToLogicalRatio),\n width: Math.round(element.rect.width * shrunkShotToLogicalRatio),\n height: Math.round(element.rect.height * shrunkShotToLogicalRatio),\n },\n };\n};\n\nexport const transformLogicalRectToScreenshotRect = (\n rect: Rect,\n shrunkShotToLogicalRatio: number,\n): Rect => {\n if (shrunkShotToLogicalRatio === 1) {\n return rect;\n }\n\n return {\n ...rect,\n left: Math.round(rect.left * shrunkShotToLogicalRatio),\n top: Math.round(rect.top * shrunkShotToLogicalRatio),\n width: Math.round(rect.width * shrunkShotToLogicalRatio),\n height: Math.round(rect.height * shrunkShotToLogicalRatio),\n };\n};\n"],"names":["commonContextParser","interfaceInstance","_opt","debug","getDebug","assert","description","uploadTestInfoToServer","interfaceSize","logicalWidth","logicalHeight","console","Number","Error","screenshotBase64","screenshotCapturedAt","Date","imgWidth","imgHeight","imageInfoOfBase64","logicalIsPortrait","screenshotIsPortrait","finalLogicalWidth","userShrinkFactor","dpr","shrunkShotToLogicalRatio","targetWidth","Math","targetHeight","resizedBase64","resizeImgBase64","ScreenshotItem","getReportFileName","tag","reportTagName","globalConfigManager","MIDSCENE_REPORT_TAG_NAME","dateTimeInFileName","dayjs","uniqueId","uuid","printReportMsg","filepath","MIDSCENE_REPORT_QUIET","logMsg","getCurrentExecutionFile","trace","error","stackTrace","pkgDir","process","stackLines","line","match","targetFileName","testFileIndex","Map","generateCacheId","fileName","taskFile","currentIndex","undefined","ifPlanLocateParamIsBbox","planLocateParam","Array","matchElementFromPlan","rect","element","generateElementByRect","matchElementFromCache","context","cacheEntry","cachePrompt","cacheable","cacheDebug","getMidsceneVersion","__VERSION__","parsePrompt","prompt","transformLogicalElementToScreenshot","transformLogicalRectToScreenshotRect"],"mappings":";;;;;;;;;AA0BO,eAAeA,oBACpBC,iBAAoC,EACpCC,IAIC;IAED,MAAMC,QAAQC,SAAS;IAEvBC,OAAOJ,mBAAmB;IAE1BE,MAAM;IACN,MAAMG,cAAcL,kBAAkB,QAAQ,QAAQ;IACtDE,MAAM;IAENA,MAAM;IACNI,uBAAuB;QACrB,SAASD;QACT,WAAWJ,KAAK,eAAe;IACjC;IACAC,MAAM;IAENA,MAAM;IACN,MAAMK,gBAAgB,MAAMP,kBAAkB,IAAI;IAClD,MAAM,EAAE,OAAOQ,YAAY,EAAE,QAAQC,aAAa,EAAE,GAAGF;IAEvD,IAAKA,cAA6C,GAAG,EACnDG,QAAQ,IAAI,CACV;IAIJ,IAAI,CAACC,OAAO,QAAQ,CAACH,iBAAiB,CAACG,OAAO,QAAQ,CAACF,gBACrD,MAAM,IAAIG,MACR,CAAC,iFAAiF,EAAEJ,aAAa,UAAU,EAAEC,eAAe;IAIhI,IAAID,gBAAgB,KAAKC,iBAAiB,GACxC,MAAM,IAAIG,MACR,CAAC,mFAAmF,EAAEJ,aAAa,UAAU,EAAEC,eAAe;IAIlIP,MAAM,CAAC,MAAM,EAAEM,aAAa,CAAC,EAAEC,eAAe;IAE9C,MAAMI,mBAAmB,MAAMb,kBAAkB,gBAAgB;IACjE,MAAMc,uBAAuBC,KAAK,GAAG;IACrCX,OAAOS,kBAAmB;IAG1BX,MAAM;IACN,MAAM,EAAE,OAAOc,QAAQ,EAAE,QAAQC,SAAS,EAAE,GAC1C,MAAMC,kBAAkBL;IAE1B,IAAI,CAACF,OAAO,QAAQ,CAACK,aAAa,CAACL,OAAO,QAAQ,CAACM,YACjD,MAAM,IAAIL,MACR,CAAC,wFAAwF,EAAEI,SAAS,UAAU,EAAEC,WAAW;IAG/H,IAAID,YAAY,KAAKC,aAAa,GAChC,MAAM,IAAIL,MACR,CAAC,0FAA0F,EAAEI,SAAS,UAAU,EAAEC,WAAW;IAGjIf,MAAM,yBAAyBc,UAAU,KAAKC;IAM9C,MAAME,oBAAoBX,eAAeC;IACzC,MAAMW,uBAAuBJ,WAAWC;IACxC,IAAII,oBAAoBb;IAExB,IAAIW,sBAAsBC,sBAAsB;QAC9ClB,MACE,CAAC,4CAA4C,EAAEM,aAAa,CAAC,EAAEC,cAAc,EAAE,EAAEU,oBAAoB,aAAa,YAAY,gBAAgB,EAAEH,SAAS,CAAC,EAAEC,UAAU,EAAE,EAAEG,uBAAuB,aAAa,YAAY,+BAA+B,CAAC;QAE5PC,oBAAoBZ;IAEtB;IAEA,MAAMa,mBAAmBrB,KAAK,sBAAsB,IAAI;IAExD,IAAI,CAACU,OAAO,QAAQ,CAACW,qBAAqBA,mBAAmB,GAC3D,MAAM,IAAIV,MACR,CAAC,wEAAwE,EAAEU,kBAAkB;IAIjG,MAAMC,MAAMP,WAAWK;IAEvBnB,MAAM,mBAAmBqB;IAEzB,MAAMC,2BAA2BD,MAAMD;IAEvCpB,MAAM,4BAA4BsB;IAElC,IAAIF,AAAqB,MAArBA,kBAAwB;QAC1B,MAAMG,cAAcC,KAAK,KAAK,CAACV,WAAWM;QAC1C,MAAMK,eAAeD,KAAK,KAAK,CAACT,YAAYK;QAE5CpB,MACE,CAAC,mCAAmC,EAAEoB,iBAAiB,YAAY,EAAEN,SAAS,CAAC,EAAEC,UAAU,YAAY,EAAEQ,YAAY,CAAC,EAAEE,aAAa,CAAC,CAAC;QAGzI,MAAMC,gBAAgB,MAAMC,gBAAgBhB,kBAAkB;YAC5D,OAAOY;YACP,QAAQE;QACV;QACA,OAAO;YACL,UAAU;gBACR,OAAOF;gBACP,QAAQE;YACV;YACA,eAAeJ;YACf,YAAYO,eAAe,MAAM,CAACF,eAAed;YACjDU;QACF;IACF;IAEA,OAAO;QACL,UAAU;YACR,OAAOR;YACP,QAAQC;QACV;QACA,eAAeM;QACf,YAAYO,eAAe,MAAM,CAACjB,kBAAkBC;QACpDU;IACF;AACF;AAEO,SAASO,kBAAkBC,MAAM,KAAK;IAC3C,MAAMC,gBAAgBC,oBAAoB,iBAAiB,CACzDC;IAEF,MAAMC,qBAAqBC,QAAQ,MAAM,CAAC;IAE1C,MAAMC,WAAWC,OAAO,SAAS,CAAC,GAAG;IACrC,OAAO,GAAGN,iBAAiBD,IAAI,CAAC,EAAEI,mBAAmB,CAAC,EAAEE,UAAU;AACpE;AAEO,SAASE,eAAeC,QAAgB;IAC7C,IAAIP,oBAAoB,qBAAqB,CAACQ,wBAC5C;IAEFC,OAAO,CAAC,gCAAgC,EAAEF,UAAU;AACtD;AAMO,SAASG,wBAAwBC,KAAc;IACpD,MAAMC,QAAQ,IAAIlC;IAClB,MAAMmC,aAAaF,SAASC,MAAM,KAAK;IACvC,MAAME,SAASC,QAAQ,GAAG,MAAM;IAChC,IAAIF,YAAY;QACd,MAAMG,aAAaH,WAAW,KAAK,CAAC;QACpC,KAAK,MAAMI,QAAQD,WACjB,IACEC,KAAK,QAAQ,CAAC,aACdA,KAAK,QAAQ,CAAC,aACdA,KAAK,QAAQ,CAAC,UACdA,KAAK,QAAQ,CAAC,QACd;YACA,MAAMC,QAAQD,KAAK,KAAK,CAAC;YACzB,IAAIC,OAAO,CAAC,EAAE,EAAE;gBACd,MAAMC,iBAAiBD,KAAK,CAAC,EAAE,CAC5B,OAAO,CAACJ,QAAQ,IAChB,IAAI,GACJ,OAAO,CAAC,OAAO;gBAClB,OAAOK;YACT;QACF;IAEJ;IACA,OAAO;AACT;AAEA,MAAMC,gBAAgB,IAAIC;AAEnB,SAASC,gBAAgBC,QAAiB;IAC/C,IAAIC,WAAWD,YAAYb;IAC3B,IAAI,CAACc,UAAU;QACbA,WAAWnB;QACX7B,QAAQ,IAAI,CACV;IAEJ;IAEA,IAAI4C,cAAc,GAAG,CAACI,WAAW;QAC/B,MAAMC,eAAeL,cAAc,GAAG,CAACI;QACvC,IAAIC,AAAiBC,WAAjBD,cACFL,cAAc,GAAG,CAACI,UAAUC,eAAe;IAE/C,OACEL,cAAc,GAAG,CAACI,UAAU;IAE9B,OAAO,GAAGA,SAAS,CAAC,EAAEJ,cAAc,GAAG,CAACI,WAAW;AACrD;AAEO,SAASG,wBACdC,eAAoC;IAEpC,OAAO,CAAC,CACNA,CAAAA,gBAAgB,IAAI,IACpBC,MAAM,OAAO,CAACD,gBAAgB,IAAI,KAClCA,AAAgC,MAAhCA,gBAAgB,IAAI,CAAC,MAAM,AAAK;AAEpC;AAEO,SAASE,qBACdF,eAAoC;IAEpC,IAAI,CAACA,iBACH;IAGF,IAAIA,gBAAgB,IAAI,EAAE;QAExB,MAAMG,OAAO;YACX,MAAMH,gBAAgB,IAAI,CAAC,EAAE;YAC7B,KAAKA,gBAAgB,IAAI,CAAC,EAAE;YAC5B,OAAOA,gBAAgB,IAAI,CAAC,EAAE,GAAGA,gBAAgB,IAAI,CAAC,EAAE,GAAG;YAC3D,QAAQA,gBAAgB,IAAI,CAAC,EAAE,GAAGA,gBAAgB,IAAI,CAAC,EAAE,GAAG;QAC9D;QAEA,MAAMI,UAAUC,sBACdF,MACA,AAAkC,YAAlC,OAAOH,gBAAgB,MAAM,GACzBA,gBAAgB,MAAM,GACtBA,gBAAgB,MAAM,EAAE,UAAU;QAExC,OAAOI;IACT;AAGF;AAEO,eAAeE,sBACpBC,OAGC,EACDC,UAA2C,EAC3CC,WAAwB,EACxBC,SAA8B;IAE9B,IAAI,CAACF,YACH;IAGF,IAAIE,AAAc,UAAdA,WAAqB,YACvBC,8BAAW,iCAAiCF;IAI9C,IAAI,CAACF,QAAQ,SAAS,EAAE,mBACtB;IAGF,IAAI,CAACA,QAAQ,iBAAiB,CAAC,uBAAuB,EAAE,YACtDI,8BACE;IAKJ,IAAI;QACF,MAAMR,OACJ,MAAMI,QAAQ,iBAAiB,CAAC,uBAAuB,CAACC;QAC1D,MAAMJ,UAA+B;YACnC,QAAQ;gBACNxC,KAAK,KAAK,CAACuC,KAAK,IAAI,GAAGA,KAAK,KAAK,GAAG;gBACpCvC,KAAK,KAAK,CAACuC,KAAK,GAAG,GAAGA,KAAK,MAAM,GAAG;aACrC;YACDA;YACA,aACE,AAAuB,YAAvB,OAAOM,cACHA,cACAA,YAAY,MAAM,IAAI;QAC9B;QAEAE,8BAAW,yBAAyBF;QACpC,OAAOL;IACT,EAAE,OAAOpB,OAAO;QACd2B,8BAAW,qCAAqC3B;QAChD;IACF;AACF;AAIO,MAAM4B,qBAAqB,IAEvBC;AAUJ,MAAMC,cAAc,CACzBC;IAKA,IAAI,AAAkB,YAAlB,OAAOA,QACT,OAAO;QACL,YAAYA;QACZ,kBAAkBjB;IACpB;IAEF,OAAO;QACL,YAAYiB,OAAO,MAAM;QACzB,kBAAkBA,OAAO,MAAM,GAC3B;YACE,QAAQA,OAAO,MAAM;YACrB,yBAAyB,CAAC,CAACA,OAAO,uBAAuB;QAC3D,IACAjB;IACN;AACF;AAEO,MAAMkB,sCAAsC,CACjDZ,SACA1C;IAEA,IAAIA,AAA6B,MAA7BA,0BACF,OAAO0C;IAGT,OAAO;QACL,GAAGA,OAAO;QACV,QAAQ;YACNxC,KAAK,KAAK,CAACwC,QAAQ,MAAM,CAAC,EAAE,GAAG1C;YAC/BE,KAAK,KAAK,CAACwC,QAAQ,MAAM,CAAC,EAAE,GAAG1C;SAChC;QACD,MAAM;YACJ,GAAG0C,QAAQ,IAAI;YACf,MAAMxC,KAAK,KAAK,CAACwC,QAAQ,IAAI,CAAC,IAAI,GAAG1C;YACrC,KAAKE,KAAK,KAAK,CAACwC,QAAQ,IAAI,CAAC,GAAG,GAAG1C;YACnC,OAAOE,KAAK,KAAK,CAACwC,QAAQ,IAAI,CAAC,KAAK,GAAG1C;YACvC,QAAQE,KAAK,KAAK,CAACwC,QAAQ,IAAI,CAAC,MAAM,GAAG1C;QAC3C;IACF;AACF;AAEO,MAAMuD,uCAAuC,CAClDd,MACAzC;IAEA,IAAIA,AAA6B,MAA7BA,0BACF,OAAOyC;IAGT,OAAO;QACL,GAAGA,IAAI;QACP,MAAMvC,KAAK,KAAK,CAACuC,KAAK,IAAI,GAAGzC;QAC7B,KAAKE,KAAK,KAAK,CAACuC,KAAK,GAAG,GAAGzC;QAC3B,OAAOE,KAAK,KAAK,CAACuC,KAAK,KAAK,GAAGzC;QAC/B,QAAQE,KAAK,KAAK,CAACuC,KAAK,MAAM,GAAGzC;IACnC;AACF"}
@@ -0,0 +1,237 @@
1
+ import { adaptBbox, pointToBbox } from "../../common.mjs";
2
+ import { getDebug } from "@midscene/shared/logger";
3
+ const debug = getDebug('auto-glm-actions');
4
+ const AUTO_GLM_COORDINATE_MAX = 1000;
5
+ function autoGLMCoordinateToBbox(x, y, width, height) {
6
+ const bbox = pointToBbox(x, y, 10);
7
+ return adaptBbox(bbox, width, height, 'auto-glm');
8
+ }
9
+ const BACK_BUTTON_NAMES = [
10
+ 'AndroidBackButton',
11
+ 'HarmonyBackButton'
12
+ ];
13
+ const HOME_BUTTON_NAMES = [
14
+ 'AndroidHomeButton',
15
+ 'HarmonyHomeButton'
16
+ ];
17
+ function findActionName(actionSpace, knownNames, defaultName) {
18
+ if (!actionSpace) return defaultName;
19
+ const match = actionSpace.find((a)=>knownNames.includes(a.name));
20
+ return match ? match.name : defaultName;
21
+ }
22
+ function transformAutoGLMAction(action, size, actionSpace) {
23
+ try {
24
+ switch(action._metadata){
25
+ case 'finish':
26
+ {
27
+ const finishAction = action;
28
+ debug('Transform finish action:', finishAction);
29
+ return [
30
+ {
31
+ type: 'Finished',
32
+ param: {},
33
+ thought: finishAction.message
34
+ }
35
+ ];
36
+ }
37
+ case 'do':
38
+ {
39
+ const doAction = action;
40
+ switch(doAction.action){
41
+ case 'Tap':
42
+ {
43
+ const tapAction = doAction;
44
+ debug('Transform Tap action:', tapAction);
45
+ const [x1, y1, x2, y2] = autoGLMCoordinateToBbox(tapAction.element[0], tapAction.element[1], size.width, size.height);
46
+ const locate = {
47
+ prompt: '',
48
+ bbox: [
49
+ x1,
50
+ y1,
51
+ x2,
52
+ y2
53
+ ]
54
+ };
55
+ return [
56
+ {
57
+ type: 'Tap',
58
+ param: {
59
+ locate
60
+ }
61
+ }
62
+ ];
63
+ }
64
+ case 'Double Tap':
65
+ {
66
+ const doubleTapAction = doAction;
67
+ debug('Transform Double Tap action:', doubleTapAction);
68
+ const [x1, y1, x2, y2] = autoGLMCoordinateToBbox(doubleTapAction.element[0], doubleTapAction.element[1], size.width, size.height);
69
+ const locate = {
70
+ prompt: '',
71
+ bbox: [
72
+ x1,
73
+ y1,
74
+ x2,
75
+ y2
76
+ ]
77
+ };
78
+ return [
79
+ {
80
+ type: 'DoubleClick',
81
+ param: {
82
+ locate
83
+ }
84
+ }
85
+ ];
86
+ }
87
+ case 'Type':
88
+ {
89
+ const typeAction = doAction;
90
+ debug('Transform Type action:', typeAction);
91
+ return [
92
+ {
93
+ type: 'Input',
94
+ param: {
95
+ value: typeAction.text
96
+ }
97
+ }
98
+ ];
99
+ }
100
+ case 'Swipe':
101
+ {
102
+ const swipeAction = doAction;
103
+ debug('Transform Swipe action:', swipeAction);
104
+ const [x1, y1, x2, y2] = autoGLMCoordinateToBbox(swipeAction.start[0], swipeAction.start[1], size.width, size.height);
105
+ const locate = {
106
+ prompt: '',
107
+ bbox: [
108
+ x1,
109
+ y1,
110
+ x2,
111
+ y2
112
+ ]
113
+ };
114
+ const deltaX = swipeAction.end[0] - swipeAction.start[0];
115
+ const deltaY = swipeAction.end[1] - swipeAction.start[1];
116
+ let direction;
117
+ let distance;
118
+ const absDeltaX = Math.abs(deltaX);
119
+ const absDeltaY = Math.abs(deltaY);
120
+ if (absDeltaY > absDeltaX) {
121
+ distance = Math.round(absDeltaY * size.height / AUTO_GLM_COORDINATE_MAX);
122
+ direction = deltaY > 0 ? 'up' : 'down';
123
+ } else {
124
+ distance = Math.round(absDeltaX * size.width / AUTO_GLM_COORDINATE_MAX);
125
+ direction = deltaX > 0 ? 'left' : 'right';
126
+ }
127
+ debug(`Calculate swipe direction: ${direction}, distance: ${distance}`);
128
+ return [
129
+ {
130
+ type: 'Scroll',
131
+ param: {
132
+ locate,
133
+ distance,
134
+ direction
135
+ },
136
+ thought: swipeAction.think || ''
137
+ }
138
+ ];
139
+ }
140
+ case 'Long Press':
141
+ {
142
+ const longPressAction = doAction;
143
+ debug('Transform Long Press action:', longPressAction);
144
+ const [x1, y1, x2, y2] = autoGLMCoordinateToBbox(longPressAction.element[0], longPressAction.element[1], size.width, size.height);
145
+ const locate = {
146
+ prompt: '',
147
+ bbox: [
148
+ x1,
149
+ y1,
150
+ x2,
151
+ y2
152
+ ]
153
+ };
154
+ return [
155
+ {
156
+ type: 'LongPress',
157
+ param: {
158
+ locate
159
+ },
160
+ thought: longPressAction.think || ''
161
+ }
162
+ ];
163
+ }
164
+ case 'Back':
165
+ {
166
+ const backAction = doAction;
167
+ debug('Transform Back action:', backAction);
168
+ return [
169
+ {
170
+ type: findActionName(actionSpace, BACK_BUTTON_NAMES, 'AndroidBackButton'),
171
+ param: {},
172
+ thought: backAction.think || ''
173
+ }
174
+ ];
175
+ }
176
+ case 'Home':
177
+ {
178
+ const homeAction = doAction;
179
+ debug('Transform Home action:', homeAction);
180
+ return [
181
+ {
182
+ type: findActionName(actionSpace, HOME_BUTTON_NAMES, 'AndroidHomeButton'),
183
+ param: {},
184
+ thought: homeAction.think || ''
185
+ }
186
+ ];
187
+ }
188
+ case 'Wait':
189
+ {
190
+ const waitAction = doAction;
191
+ debug('Transform Wait action:', waitAction);
192
+ return [
193
+ {
194
+ type: 'Sleep',
195
+ param: {
196
+ timeMs: waitAction.durationMs
197
+ },
198
+ thought: waitAction.think || ''
199
+ }
200
+ ];
201
+ }
202
+ case 'Launch':
203
+ {
204
+ const launchAction = doAction;
205
+ debug('Transform Launch action:', launchAction);
206
+ return [
207
+ {
208
+ type: 'Launch',
209
+ param: launchAction.app,
210
+ thought: launchAction.think || ''
211
+ }
212
+ ];
213
+ }
214
+ case 'Interact':
215
+ throw new Error('Action "Interact" from auto-glm is not supported in the current implementation.');
216
+ case 'Call_API':
217
+ throw new Error('Action "Call_API" from auto-glm is not supported in the current implementation.');
218
+ case 'Take_over':
219
+ throw new Error('Action "Take_over" from auto-glm is not supported in the current implementation.');
220
+ case 'Note':
221
+ throw new Error('Action "Note" from auto-glm is not supported in the current implementation.');
222
+ default:
223
+ throw new Error(`Unknown do() action type: ${doAction.action}`);
224
+ }
225
+ }
226
+ default:
227
+ throw new Error(`Unknown action metadata: ${action._metadata}`);
228
+ }
229
+ } catch (error) {
230
+ const errorMessage = error instanceof Error ? error.message : String(error);
231
+ debug('Transform error:', errorMessage);
232
+ throw new Error(`Failed to transform action: ${errorMessage}`);
233
+ }
234
+ }
235
+ export { transformAutoGLMAction };
236
+
237
+ //# sourceMappingURL=actions.mjs.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"ai-model/auto-glm/actions.mjs","sources":["../../../../src/ai-model/auto-glm/actions.ts"],"sourcesContent":["import { adaptBbox, pointToBbox } from '@/common';\nimport type { DeviceAction } from '@/device';\nimport type { PlanningAction } from '@/types';\nimport { getDebug } from '@midscene/shared/logger';\n\nconst debug = getDebug('auto-glm-actions');\n\n/**\n * Auto-GLM coordinate system range: [0, AUTO_GLM_COORDINATE_MAX]\n */\nconst AUTO_GLM_COORDINATE_MAX = 1000;\n\n/**\n * Convert auto-glm coordinate [0,1000] to bbox in pixel coordinates\n */\nfunction autoGLMCoordinateToBbox(\n x: number,\n y: number,\n width: number,\n height: number,\n): [number, number, number, number] {\n const bbox = pointToBbox(x, y, 10);\n return adaptBbox(bbox, width, height, 'auto-glm');\n}\n\nexport interface BaseAction {\n _metadata: string;\n think?: string;\n}\n\nexport interface TapAction extends BaseAction {\n _metadata: 'do';\n action: 'Tap';\n element: [number, number];\n}\n\nexport interface DoubleTapAction extends BaseAction {\n _metadata: 'do';\n action: 'Double Tap';\n element: [number, number];\n}\n\nexport interface TypeAction extends BaseAction {\n _metadata: 'do';\n action: 'Type';\n text: string;\n}\n\nexport interface SwipeAction extends BaseAction {\n _metadata: 'do';\n action: 'Swipe';\n start: [number, number];\n end: [number, number];\n}\n\nexport interface LongPressAction extends BaseAction {\n _metadata: 'do';\n action: 'Long Press';\n element: [number, number];\n}\n\nexport interface LaunchAction extends BaseAction {\n _metadata: 'do';\n action: 'Launch';\n app: string;\n}\n\nexport interface BackAction extends BaseAction {\n _metadata: 'do';\n action: 'Back';\n}\n\nexport interface HomeAction extends BaseAction {\n _metadata: 'do';\n action: 'Home';\n}\n\nexport interface WaitAction extends BaseAction {\n _metadata: 'do';\n action: 'Wait';\n durationMs: number;\n}\n\nexport interface InteractAction extends BaseAction {\n _metadata: 'do';\n action: 'Interact';\n}\n\nexport interface CallAPIAction extends BaseAction {\n _metadata: 'do';\n action: 'Call_API';\n instruction: string;\n}\n\nexport interface TakeoverAction extends BaseAction {\n _metadata: 'do';\n action: 'Take_over';\n message: string;\n}\n\nexport interface NoteAction extends BaseAction {\n _metadata: 'do';\n action: 'Note';\n message: string;\n}\n\nexport interface FinishAction extends BaseAction {\n _metadata: 'finish';\n message: string;\n}\n\nexport type ParsedAction =\n | TapAction\n | DoubleTapAction\n | TypeAction\n | SwipeAction\n | LongPressAction\n | LaunchAction\n | BackAction\n | HomeAction\n | WaitAction\n | InteractAction\n | CallAPIAction\n | TakeoverAction\n | NoteAction\n | FinishAction;\n\nconst BACK_BUTTON_NAMES = ['AndroidBackButton', 'HarmonyBackButton'];\nconst HOME_BUTTON_NAMES = ['AndroidHomeButton', 'HarmonyHomeButton'];\n\n/**\n * Find the action name in actionSpace that matches one of the known names.\n * Falls back to defaultName if no match found or actionSpace is not provided.\n */\nfunction findActionName(\n actionSpace: DeviceAction[] | undefined,\n knownNames: string[],\n defaultName: string,\n): string {\n if (!actionSpace) return defaultName;\n const match = actionSpace.find((a) => knownNames.includes(a.name));\n return match ? match.name : defaultName;\n}\n\nexport function transformAutoGLMAction(\n action: ParsedAction,\n size: { width: number; height: number },\n actionSpace?: DeviceAction[],\n): PlanningAction[] {\n try {\n switch (action._metadata) {\n case 'finish': {\n const finishAction = action as FinishAction;\n debug('Transform finish action:', finishAction);\n return [\n {\n type: 'Finished',\n param: {},\n thought: finishAction.message,\n },\n ];\n }\n case 'do': {\n const doAction = action as\n | TapAction\n | DoubleTapAction\n | TypeAction\n | SwipeAction\n | LongPressAction\n | LaunchAction\n | BackAction\n | HomeAction\n | WaitAction\n | InteractAction\n | CallAPIAction\n | TakeoverAction\n | NoteAction;\n\n switch ((doAction as any).action) {\n case 'Tap': {\n const tapAction = doAction as TapAction;\n debug('Transform Tap action:', tapAction);\n const [x1, y1, x2, y2] = autoGLMCoordinateToBbox(\n tapAction.element[0],\n tapAction.element[1],\n size.width,\n size.height,\n );\n\n const locate: {\n prompt: string;\n bbox: [number, number, number, number];\n } = {\n prompt: '',\n bbox: [x1, y1, x2, y2],\n };\n\n return [\n {\n type: 'Tap',\n param: {\n locate,\n },\n },\n ];\n }\n case 'Double Tap': {\n const doubleTapAction = doAction as DoubleTapAction;\n debug('Transform Double Tap action:', doubleTapAction);\n const [x1, y1, x2, y2] = autoGLMCoordinateToBbox(\n doubleTapAction.element[0],\n doubleTapAction.element[1],\n size.width,\n size.height,\n );\n\n const locate: {\n prompt: string;\n bbox: [number, number, number, number];\n } = {\n prompt: '',\n bbox: [x1, y1, x2, y2],\n };\n\n return [\n {\n type: 'DoubleClick',\n param: {\n locate,\n },\n },\n ];\n }\n case 'Type': {\n const typeAction = doAction as TypeAction;\n debug('Transform Type action:', typeAction);\n\n return [\n {\n type: 'Input',\n param: {\n value: typeAction.text,\n },\n },\n ];\n }\n case 'Swipe': {\n const swipeAction = doAction as SwipeAction;\n debug('Transform Swipe action:', swipeAction);\n\n // Calculate locate using start coordinate\n const [x1, y1, x2, y2] = autoGLMCoordinateToBbox(\n swipeAction.start[0],\n swipeAction.start[1],\n size.width,\n size.height,\n );\n\n const locate: {\n prompt: string;\n bbox: [number, number, number, number];\n } = {\n prompt: '',\n bbox: [x1, y1, x2, y2],\n };\n\n // Calculate horizontal and vertical delta in [0,AUTO_GLM_COORDINATE_MAX] coordinate system\n const deltaX = swipeAction.end[0] - swipeAction.start[0];\n const deltaY = swipeAction.end[1] - swipeAction.start[1];\n\n // Determine direction and distance\n let direction: 'up' | 'down' | 'left' | 'right';\n let distance: number;\n\n const absDeltaX = Math.abs(deltaX);\n const absDeltaY = Math.abs(deltaY);\n\n if (absDeltaY > absDeltaX) {\n // Vertical scroll\n distance = Math.round(\n (absDeltaY * size.height) / AUTO_GLM_COORDINATE_MAX,\n );\n direction = deltaY > 0 ? 'up' : 'down';\n } else {\n // Horizontal scroll\n distance = Math.round(\n (absDeltaX * size.width) / AUTO_GLM_COORDINATE_MAX,\n );\n direction = deltaX > 0 ? 'left' : 'right';\n }\n\n debug(\n `Calculate swipe direction: ${direction}, distance: ${distance}`,\n );\n\n return [\n {\n type: 'Scroll',\n param: {\n locate,\n // The scrolling direction here all refers to which direction of the page's content will appear on the screen.\n distance,\n direction,\n },\n thought: swipeAction.think || '',\n },\n ];\n }\n case 'Long Press': {\n const longPressAction = doAction as LongPressAction;\n debug('Transform Long Press action:', longPressAction);\n const [x1, y1, x2, y2] = autoGLMCoordinateToBbox(\n longPressAction.element[0],\n longPressAction.element[1],\n size.width,\n size.height,\n );\n\n const locate: {\n prompt: string;\n bbox: [number, number, number, number];\n } = {\n prompt: '',\n bbox: [x1, y1, x2, y2],\n };\n\n return [\n {\n type: 'LongPress',\n param: {\n locate,\n },\n thought: longPressAction.think || '',\n },\n ];\n }\n case 'Back': {\n const backAction = doAction as BackAction;\n debug('Transform Back action:', backAction);\n return [\n {\n type: findActionName(\n actionSpace,\n BACK_BUTTON_NAMES,\n 'AndroidBackButton',\n ),\n param: {},\n thought: backAction.think || '',\n },\n ];\n }\n case 'Home': {\n const homeAction = doAction as HomeAction;\n debug('Transform Home action:', homeAction);\n return [\n {\n type: findActionName(\n actionSpace,\n HOME_BUTTON_NAMES,\n 'AndroidHomeButton',\n ),\n param: {},\n thought: homeAction.think || '',\n },\n ];\n }\n case 'Wait': {\n const waitAction = doAction as WaitAction;\n debug('Transform Wait action:', waitAction);\n return [\n {\n type: 'Sleep',\n param: {\n timeMs: waitAction.durationMs,\n },\n thought: waitAction.think || '',\n },\n ];\n }\n case 'Launch': {\n const launchAction = doAction as LaunchAction;\n debug('Transform Launch action:', launchAction);\n return [\n {\n type: 'Launch',\n param: launchAction.app,\n thought: launchAction.think || '',\n },\n ];\n }\n case 'Interact': {\n throw new Error(\n `Action \"Interact\" from auto-glm is not supported in the current implementation.`,\n );\n }\n case 'Call_API': {\n throw new Error(\n `Action \"Call_API\" from auto-glm is not supported in the current implementation.`,\n );\n }\n case 'Take_over': {\n throw new Error(\n `Action \"Take_over\" from auto-glm is not supported in the current implementation.`,\n );\n }\n case 'Note': {\n throw new Error(\n `Action \"Note\" from auto-glm is not supported in the current implementation.`,\n );\n }\n default:\n throw new Error(\n `Unknown do() action type: ${(doAction as any).action}`,\n );\n }\n }\n default:\n throw new Error(\n `Unknown action metadata: ${(action as any)._metadata}`,\n );\n }\n } catch (error) {\n const errorMessage = error instanceof Error ? error.message : String(error);\n debug('Transform error:', errorMessage);\n throw new Error(`Failed to transform action: ${errorMessage}`);\n }\n}\n"],"names":["debug","getDebug","AUTO_GLM_COORDINATE_MAX","autoGLMCoordinateToBbox","x","y","width","height","bbox","pointToBbox","adaptBbox","BACK_BUTTON_NAMES","HOME_BUTTON_NAMES","findActionName","actionSpace","knownNames","defaultName","match","a","transformAutoGLMAction","action","size","finishAction","doAction","tapAction","x1","y1","x2","y2","locate","doubleTapAction","typeAction","swipeAction","deltaX","deltaY","direction","distance","absDeltaX","Math","absDeltaY","longPressAction","backAction","homeAction","waitAction","launchAction","Error","error","errorMessage","String"],"mappings":";;AAKA,MAAMA,QAAQC,SAAS;AAKvB,MAAMC,0BAA0B;AAKhC,SAASC,wBACPC,CAAS,EACTC,CAAS,EACTC,KAAa,EACbC,MAAc;IAEd,MAAMC,OAAOC,YAAYL,GAAGC,GAAG;IAC/B,OAAOK,UAAUF,MAAMF,OAAOC,QAAQ;AACxC;AAwGA,MAAMI,oBAAoB;IAAC;IAAqB;CAAoB;AACpE,MAAMC,oBAAoB;IAAC;IAAqB;CAAoB;AAMpE,SAASC,eACPC,WAAuC,EACvCC,UAAoB,EACpBC,WAAmB;IAEnB,IAAI,CAACF,aAAa,OAAOE;IACzB,MAAMC,QAAQH,YAAY,IAAI,CAAC,CAACI,IAAMH,WAAW,QAAQ,CAACG,EAAE,IAAI;IAChE,OAAOD,QAAQA,MAAM,IAAI,GAAGD;AAC9B;AAEO,SAASG,uBACdC,MAAoB,EACpBC,IAAuC,EACvCP,WAA4B;IAE5B,IAAI;QACF,OAAQM,OAAO,SAAS;YACtB,KAAK;gBAAU;oBACb,MAAME,eAAeF;oBACrBpB,MAAM,4BAA4BsB;oBAClC,OAAO;wBACL;4BACE,MAAM;4BACN,OAAO,CAAC;4BACR,SAASA,aAAa,OAAO;wBAC/B;qBACD;gBACH;YACA,KAAK;gBAAM;oBACT,MAAMC,WAAWH;oBAejB,OAASG,SAAiB,MAAM;wBAC9B,KAAK;4BAAO;gCACV,MAAMC,YAAYD;gCAClBvB,MAAM,yBAAyBwB;gCAC/B,MAAM,CAACC,IAAIC,IAAIC,IAAIC,GAAG,GAAGzB,wBACvBqB,UAAU,OAAO,CAAC,EAAE,EACpBA,UAAU,OAAO,CAAC,EAAE,EACpBH,KAAK,KAAK,EACVA,KAAK,MAAM;gCAGb,MAAMQ,SAGF;oCACF,QAAQ;oCACR,MAAM;wCAACJ;wCAAIC;wCAAIC;wCAAIC;qCAAG;gCACxB;gCAEA,OAAO;oCACL;wCACE,MAAM;wCACN,OAAO;4CACLC;wCACF;oCACF;iCACD;4BACH;wBACA,KAAK;4BAAc;gCACjB,MAAMC,kBAAkBP;gCACxBvB,MAAM,gCAAgC8B;gCACtC,MAAM,CAACL,IAAIC,IAAIC,IAAIC,GAAG,GAAGzB,wBACvB2B,gBAAgB,OAAO,CAAC,EAAE,EAC1BA,gBAAgB,OAAO,CAAC,EAAE,EAC1BT,KAAK,KAAK,EACVA,KAAK,MAAM;gCAGb,MAAMQ,SAGF;oCACF,QAAQ;oCACR,MAAM;wCAACJ;wCAAIC;wCAAIC;wCAAIC;qCAAG;gCACxB;gCAEA,OAAO;oCACL;wCACE,MAAM;wCACN,OAAO;4CACLC;wCACF;oCACF;iCACD;4BACH;wBACA,KAAK;4BAAQ;gCACX,MAAME,aAAaR;gCACnBvB,MAAM,0BAA0B+B;gCAEhC,OAAO;oCACL;wCACE,MAAM;wCACN,OAAO;4CACL,OAAOA,WAAW,IAAI;wCACxB;oCACF;iCACD;4BACH;wBACA,KAAK;4BAAS;gCACZ,MAAMC,cAAcT;gCACpBvB,MAAM,2BAA2BgC;gCAGjC,MAAM,CAACP,IAAIC,IAAIC,IAAIC,GAAG,GAAGzB,wBACvB6B,YAAY,KAAK,CAAC,EAAE,EACpBA,YAAY,KAAK,CAAC,EAAE,EACpBX,KAAK,KAAK,EACVA,KAAK,MAAM;gCAGb,MAAMQ,SAGF;oCACF,QAAQ;oCACR,MAAM;wCAACJ;wCAAIC;wCAAIC;wCAAIC;qCAAG;gCACxB;gCAGA,MAAMK,SAASD,YAAY,GAAG,CAAC,EAAE,GAAGA,YAAY,KAAK,CAAC,EAAE;gCACxD,MAAME,SAASF,YAAY,GAAG,CAAC,EAAE,GAAGA,YAAY,KAAK,CAAC,EAAE;gCAGxD,IAAIG;gCACJ,IAAIC;gCAEJ,MAAMC,YAAYC,KAAK,GAAG,CAACL;gCAC3B,MAAMM,YAAYD,KAAK,GAAG,CAACJ;gCAE3B,IAAIK,YAAYF,WAAW;oCAEzBD,WAAWE,KAAK,KAAK,CAClBC,YAAYlB,KAAK,MAAM,GAAInB;oCAE9BiC,YAAYD,SAAS,IAAI,OAAO;gCAClC,OAAO;oCAELE,WAAWE,KAAK,KAAK,CAClBD,YAAYhB,KAAK,KAAK,GAAInB;oCAE7BiC,YAAYF,SAAS,IAAI,SAAS;gCACpC;gCAEAjC,MACE,CAAC,2BAA2B,EAAEmC,UAAU,YAAY,EAAEC,UAAU;gCAGlE,OAAO;oCACL;wCACE,MAAM;wCACN,OAAO;4CACLP;4CAEAO;4CACAD;wCACF;wCACA,SAASH,YAAY,KAAK,IAAI;oCAChC;iCACD;4BACH;wBACA,KAAK;4BAAc;gCACjB,MAAMQ,kBAAkBjB;gCACxBvB,MAAM,gCAAgCwC;gCACtC,MAAM,CAACf,IAAIC,IAAIC,IAAIC,GAAG,GAAGzB,wBACvBqC,gBAAgB,OAAO,CAAC,EAAE,EAC1BA,gBAAgB,OAAO,CAAC,EAAE,EAC1BnB,KAAK,KAAK,EACVA,KAAK,MAAM;gCAGb,MAAMQ,SAGF;oCACF,QAAQ;oCACR,MAAM;wCAACJ;wCAAIC;wCAAIC;wCAAIC;qCAAG;gCACxB;gCAEA,OAAO;oCACL;wCACE,MAAM;wCACN,OAAO;4CACLC;wCACF;wCACA,SAASW,gBAAgB,KAAK,IAAI;oCACpC;iCACD;4BACH;wBACA,KAAK;4BAAQ;gCACX,MAAMC,aAAalB;gCACnBvB,MAAM,0BAA0ByC;gCAChC,OAAO;oCACL;wCACE,MAAM5B,eACJC,aACAH,mBACA;wCAEF,OAAO,CAAC;wCACR,SAAS8B,WAAW,KAAK,IAAI;oCAC/B;iCACD;4BACH;wBACA,KAAK;4BAAQ;gCACX,MAAMC,aAAanB;gCACnBvB,MAAM,0BAA0B0C;gCAChC,OAAO;oCACL;wCACE,MAAM7B,eACJC,aACAF,mBACA;wCAEF,OAAO,CAAC;wCACR,SAAS8B,WAAW,KAAK,IAAI;oCAC/B;iCACD;4BACH;wBACA,KAAK;4BAAQ;gCACX,MAAMC,aAAapB;gCACnBvB,MAAM,0BAA0B2C;gCAChC,OAAO;oCACL;wCACE,MAAM;wCACN,OAAO;4CACL,QAAQA,WAAW,UAAU;wCAC/B;wCACA,SAASA,WAAW,KAAK,IAAI;oCAC/B;iCACD;4BACH;wBACA,KAAK;4BAAU;gCACb,MAAMC,eAAerB;gCACrBvB,MAAM,4BAA4B4C;gCAClC,OAAO;oCACL;wCACE,MAAM;wCACN,OAAOA,aAAa,GAAG;wCACvB,SAASA,aAAa,KAAK,IAAI;oCACjC;iCACD;4BACH;wBACA,KAAK;4BACH,MAAM,IAAIC,MACR;wBAGJ,KAAK;4BACH,MAAM,IAAIA,MACR;wBAGJ,KAAK;4BACH,MAAM,IAAIA,MACR;wBAGJ,KAAK;4BACH,MAAM,IAAIA,MACR;wBAGJ;4BACE,MAAM,IAAIA,MACR,CAAC,0BAA0B,EAAGtB,SAAiB,MAAM,EAAE;oBAE7D;gBACF;YACA;gBACE,MAAM,IAAIsB,MACR,CAAC,yBAAyB,EAAGzB,OAAe,SAAS,EAAE;QAE7D;IACF,EAAE,OAAO0B,OAAO;QACd,MAAMC,eAAeD,iBAAiBD,QAAQC,MAAM,OAAO,GAAGE,OAAOF;QACrE9C,MAAM,oBAAoB+C;QAC1B,MAAM,IAAIF,MAAM,CAAC,4BAA4B,EAAEE,cAAc;IAC/D;AACF"}
@@ -0,0 +1,6 @@
1
+ import { getAutoGLMLocatePrompt, getAutoGLMPlanPrompt } from "./prompt.mjs";
2
+ import { parseAction, parseAutoGLMLocateResponse, parseAutoGLMResponse } from "./parser.mjs";
3
+ import { autoGLMPlanning } from "./planning.mjs";
4
+ import { transformAutoGLMAction } from "./actions.mjs";
5
+ import { isAutoGLM, isUITars } from "./util.mjs";
6
+ export { autoGLMPlanning, getAutoGLMLocatePrompt, getAutoGLMPlanPrompt, isAutoGLM, isUITars, parseAction, parseAutoGLMLocateResponse, parseAutoGLMResponse, transformAutoGLMAction };