@asterxsk/kiln 0.4.0 → 0.4.2

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 (259) hide show
  1. package/README.md +155 -155
  2. package/agent/AGENTS.md +67 -67
  3. package/agent/README.md +5 -5
  4. package/agent/extensions/ask-user/index.ts +418 -418
  5. package/agent/extensions/ask-user/install.ps1 +27 -27
  6. package/agent/extensions/ask-user/install.sh +24 -24
  7. package/agent/extensions/ask-user/package-lock.json +769 -769
  8. package/agent/extensions/ask-user/package.json +19 -19
  9. package/agent/extensions/ask-user/prompt.ts +45 -45
  10. package/agent/extensions/ask-user/tsconfig.json +7 -7
  11. package/agent/extensions/background-terminals/docs/implementation-guide.md +942 -942
  12. package/agent/extensions/background-terminals/index.ts +627 -627
  13. package/agent/extensions/background-terminals/install.ps1 +27 -27
  14. package/agent/extensions/background-terminals/install.sh +24 -24
  15. package/agent/extensions/background-terminals/manager.test.ts +735 -735
  16. package/agent/extensions/background-terminals/output.test.ts +109 -109
  17. package/agent/extensions/background-terminals/package-lock.json +769 -769
  18. package/agent/extensions/background-terminals/package.json +17 -17
  19. package/agent/extensions/background-terminals/prompt.test.ts +125 -125
  20. package/agent/extensions/background-terminals/ps.test.ts +82 -82
  21. package/agent/extensions/background-terminals/result-delivery.test.ts +44 -44
  22. package/agent/extensions/background-terminals/src/domain.ts +87 -87
  23. package/agent/extensions/background-terminals/src/manager.ts +907 -907
  24. package/agent/extensions/background-terminals/src/output.ts +84 -84
  25. package/agent/extensions/background-terminals/src/prompt.ts +142 -142
  26. package/agent/extensions/background-terminals/src/result-delivery.ts +27 -27
  27. package/agent/extensions/background-terminals/src/runtime.ts +36 -36
  28. package/agent/extensions/background-terminals/src/ui/output-view.ts +79 -79
  29. package/agent/extensions/background-terminals/src/ui/ps.ts +621 -621
  30. package/agent/extensions/background-terminals/tsconfig.json +7 -7
  31. package/agent/extensions/destructive/README.md +31 -31
  32. package/agent/extensions/destructive/index.ts +88 -88
  33. package/agent/extensions/destructive/install.ps1 +27 -27
  34. package/agent/extensions/destructive/install.sh +24 -24
  35. package/agent/extensions/file-search/index.spec.ts +443 -443
  36. package/agent/extensions/file-search/index.ts +459 -459
  37. package/agent/extensions/file-search/install.ps1 +27 -27
  38. package/agent/extensions/file-search/install.sh +24 -24
  39. package/agent/extensions/file-search/package-lock.json +2253 -2253
  40. package/agent/extensions/file-search/package.json +23 -23
  41. package/agent/extensions/file-search/src/args.ts +122 -122
  42. package/agent/extensions/file-search/src/binaries.ts +422 -422
  43. package/agent/extensions/file-search/src/output.ts +126 -126
  44. package/agent/extensions/file-search/src/process.ts +146 -146
  45. package/agent/extensions/file-search/src/prompt.ts +52 -52
  46. package/agent/extensions/file-search/tsconfig.json +7 -7
  47. package/agent/extensions/goal/README.md +50 -50
  48. package/agent/extensions/goal/index.ts +155 -155
  49. package/agent/extensions/goal/install.ps1 +27 -27
  50. package/agent/extensions/goal/install.sh +24 -24
  51. package/agent/extensions/kiln-update/README.md +38 -25
  52. package/agent/extensions/kiln-update/index.ts +177 -99
  53. package/agent/extensions/modelconf/PLAN.md +915 -915
  54. package/agent/extensions/modelconf/README.md +66 -66
  55. package/agent/extensions/modelconf/index.ts +296 -296
  56. package/agent/extensions/modelconf/install.ps1 +27 -27
  57. package/agent/extensions/modelconf/install.sh +24 -24
  58. package/agent/extensions/modelconf/package-lock.json +1809 -1809
  59. package/agent/extensions/modelconf/package.json +13 -13
  60. package/agent/extensions/modelconf/src/fuzzy.ts +26 -26
  61. package/agent/extensions/modelconf/src/glob.ts +13 -13
  62. package/agent/extensions/modelconf/src/persistence.test.ts +46 -46
  63. package/agent/extensions/modelconf/src/persistence.ts +164 -164
  64. package/agent/extensions/modelconf/src/ui/ModelConfView.test.ts +75 -75
  65. package/agent/extensions/modelconf/src/ui/ModelConfView.ts +1101 -1101
  66. package/agent/extensions/modelconf/tsconfig.json +15 -15
  67. package/agent/extensions/pi-web-access/CHANGELOG.md +690 -690
  68. package/agent/extensions/pi-web-access/LICENSE +21 -21
  69. package/agent/extensions/pi-web-access/README.md +470 -470
  70. package/agent/extensions/pi-web-access/SECURITY.md +5 -5
  71. package/agent/extensions/pi-web-access/activity.ts +101 -101
  72. package/agent/extensions/pi-web-access/auth-fetch.ts +148 -148
  73. package/agent/extensions/pi-web-access/brightdata-unlocker.ts +272 -272
  74. package/agent/extensions/pi-web-access/chrome-cookies.ts +669 -669
  75. package/agent/extensions/pi-web-access/content-find.ts +139 -139
  76. package/agent/extensions/pi-web-access/credential-source.ts +191 -191
  77. package/agent/extensions/pi-web-access/data-uri-sanitize.ts +406 -406
  78. package/agent/extensions/pi-web-access/datalab-pdf-extract.ts +568 -568
  79. package/agent/extensions/pi-web-access/declared-web-links.ts +173 -173
  80. package/agent/extensions/pi-web-access/evidence/CONTRACT-EVIDENCE.md +496 -496
  81. package/agent/extensions/pi-web-access/evidence/contract-probe.mjs +140 -140
  82. package/agent/extensions/pi-web-access/exa.ts +526 -526
  83. package/agent/extensions/pi-web-access/extract.ts +1196 -1196
  84. package/agent/extensions/pi-web-access/feature-config.ts +29 -29
  85. package/agent/extensions/pi-web-access/fetch-params.ts +111 -111
  86. package/agent/extensions/pi-web-access/gemini-adc.ts +298 -298
  87. package/agent/extensions/pi-web-access/gemini-api.ts +353 -353
  88. package/agent/extensions/pi-web-access/gemini-pdf-extract.ts +108 -108
  89. package/agent/extensions/pi-web-access/gemini-search.ts +21 -21
  90. package/agent/extensions/pi-web-access/gemini-url-context.ts +128 -128
  91. package/agent/extensions/pi-web-access/gemini-web-config.ts +101 -101
  92. package/agent/extensions/pi-web-access/gemini-web.ts +487 -487
  93. package/agent/extensions/pi-web-access/github-api.ts +197 -197
  94. package/agent/extensions/pi-web-access/github-extract.ts +746 -746
  95. package/agent/extensions/pi-web-access/github-issue-pr.ts +700 -700
  96. package/agent/extensions/pi-web-access/index.ts +1737 -1737
  97. package/agent/extensions/pi-web-access/package-lock.json +5808 -5808
  98. package/agent/extensions/pi-web-access/package.json +64 -64
  99. package/agent/extensions/pi-web-access/page-query.ts +96 -96
  100. package/agent/extensions/pi-web-access/pdf-extract.ts +409 -409
  101. package/agent/extensions/pi-web-access/promise-try.d.ts +7 -7
  102. package/agent/extensions/pi-web-access/query-rewrite.ts +51 -51
  103. package/agent/extensions/pi-web-access/render-search-error.ts +170 -170
  104. package/agent/extensions/pi-web-access/rsc-extract.ts +338 -338
  105. package/agent/extensions/pi-web-access/search-types.ts +20 -20
  106. package/agent/extensions/pi-web-access/source-check.ts +282 -282
  107. package/agent/extensions/pi-web-access/ssrf-protection.ts +526 -526
  108. package/agent/extensions/pi-web-access/storage.ts +521 -521
  109. package/agent/extensions/pi-web-access/summary-model-scope.ts +125 -125
  110. package/agent/extensions/pi-web-access/test/auth-fetch.test.mjs +208 -208
  111. package/agent/extensions/pi-web-access/test/brightdata-unlocker.test.mjs +840 -840
  112. package/agent/extensions/pi-web-access/test/chrome-cookie-extraction.test.mjs +441 -441
  113. package/agent/extensions/pi-web-access/test/config-path.test.mjs +283 -283
  114. package/agent/extensions/pi-web-access/test/content-find.test.mjs +25 -25
  115. package/agent/extensions/pi-web-access/test/credential-source.test.mjs +118 -118
  116. package/agent/extensions/pi-web-access/test/data-uri-sanitize.test.mjs +210 -210
  117. package/agent/extensions/pi-web-access/test/datalab-pdf-extract.test.mjs +552 -552
  118. package/agent/extensions/pi-web-access/test/declared-web-links.test.mjs +212 -212
  119. package/agent/extensions/pi-web-access/test/fetch-answer-storage.test.mjs +40 -40
  120. package/agent/extensions/pi-web-access/test/fetch-cache-storage.test.mjs +334 -334
  121. package/agent/extensions/pi-web-access/test/fetch-content-domain-policy.test.mjs +95 -95
  122. package/agent/extensions/pi-web-access/test/fetch-modes.test.mjs +53 -53
  123. package/agent/extensions/pi-web-access/test/fetch-not-found-guidance.test.mjs +92 -92
  124. package/agent/extensions/pi-web-access/test/fetch-params.test.mjs +86 -86
  125. package/agent/extensions/pi-web-access/test/fetch-render-call.test.mjs +34 -34
  126. package/agent/extensions/pi-web-access/test/fetch-routing.test.mjs +173 -173
  127. package/agent/extensions/pi-web-access/test/gemini-adc-auth.test.mjs +257 -257
  128. package/agent/extensions/pi-web-access/test/gemini-api-transport.test.mjs +170 -170
  129. package/agent/extensions/pi-web-access/test/gemini-pdf-extract.test.mjs +133 -133
  130. package/agent/extensions/pi-web-access/test/gemini-web-cookie-opt-in.test.mjs +178 -178
  131. package/agent/extensions/pi-web-access/test/gemini-web-header-overflow.test.mjs +148 -148
  132. package/agent/extensions/pi-web-access/test/get-search-content.test.mjs +223 -223
  133. package/agent/extensions/pi-web-access/test/github-extract.test.mjs +378 -378
  134. package/agent/extensions/pi-web-access/test/github-issue-pr.test.mjs +565 -565
  135. package/agent/extensions/pi-web-access/test/inline-content-config.test.mjs +99 -99
  136. package/agent/extensions/pi-web-access/test/lazy-extract-load.test.mjs +118 -118
  137. package/agent/extensions/pi-web-access/test/local-video-oversize.test.mjs +52 -52
  138. package/agent/extensions/pi-web-access/test/package-typebox-dependency.test.mjs +50 -50
  139. package/agent/extensions/pi-web-access/test/page-query.test.mjs +51 -51
  140. package/agent/extensions/pi-web-access/test/pdf-config.test.mjs +140 -140
  141. package/agent/extensions/pi-web-access/test/pdf-extract.test.mjs +500 -500
  142. package/agent/extensions/pi-web-access/test/proxy-transport.test.mjs +286 -286
  143. package/agent/extensions/pi-web-access/test/query-rewrite.test.mjs +52 -52
  144. package/agent/extensions/pi-web-access/test/rsc-fallback.test.mjs +102 -102
  145. package/agent/extensions/pi-web-access/test/search-error-render.test.mjs +152 -152
  146. package/agent/extensions/pi-web-access/test/search-providers.test.mjs +274 -274
  147. package/agent/extensions/pi-web-access/test/source-check.test.mjs +179 -179
  148. package/agent/extensions/pi-web-access/test/ssrf-allow-ranges-config.test.mjs +205 -205
  149. package/agent/extensions/pi-web-access/test/ssrf-protection.test.mjs +456 -456
  150. package/agent/extensions/pi-web-access/test/summary-model-scope.test.mjs +106 -106
  151. package/agent/extensions/pi-web-access/test/tool-registration-config.test.mjs +182 -182
  152. package/agent/extensions/pi-web-access/test/web-search-answer-render.test.mjs +66 -66
  153. package/agent/extensions/pi-web-access/test/youtube-extract-errors.test.mjs +64 -64
  154. package/agent/extensions/pi-web-access/tsconfig.json +11 -11
  155. package/agent/extensions/pi-web-access/utils.ts +451 -451
  156. package/agent/extensions/pi-web-access/video-extract.ts +392 -392
  157. package/agent/extensions/pi-web-access/youtube-extract.ts +328 -328
  158. package/agent/extensions/shared/activity-status.ts +31 -31
  159. package/agent/extensions/shared/child-session.test.ts +270 -270
  160. package/agent/extensions/shared/child-session.ts +148 -148
  161. package/agent/extensions/shared/context-utilization.test.ts +48 -48
  162. package/agent/extensions/shared/context-utilization.ts +47 -47
  163. package/agent/extensions/shared/dashboard-state.ts +99 -99
  164. package/agent/extensions/shared/install.ps1 +27 -27
  165. package/agent/extensions/shared/install.sh +24 -24
  166. package/agent/extensions/shared/tool-call-timeout.test.ts +117 -117
  167. package/agent/extensions/shared/tool-call-timeout.ts +104 -104
  168. package/agent/extensions/skillsconf/README.md +74 -74
  169. package/agent/extensions/skillsconf/index.ts +92 -92
  170. package/agent/extensions/skillsconf/install.ps1 +27 -27
  171. package/agent/extensions/skillsconf/install.sh +24 -24
  172. package/agent/extensions/skillsconf/package-lock.json +1809 -1809
  173. package/agent/extensions/skillsconf/package.json +18 -18
  174. package/agent/extensions/skillsconf/src/delete-skill.test.ts +64 -64
  175. package/agent/extensions/skillsconf/src/delete-skill.ts +45 -45
  176. package/agent/extensions/skillsconf/src/filter.test.ts +80 -80
  177. package/agent/extensions/skillsconf/src/filter.ts +74 -74
  178. package/agent/extensions/skillsconf/src/fuzzy.ts +26 -26
  179. package/agent/extensions/skillsconf/src/persistence.test.ts +72 -72
  180. package/agent/extensions/skillsconf/src/persistence.ts +169 -169
  181. package/agent/extensions/skillsconf/src/ui/SkillConfView.test.ts +337 -337
  182. package/agent/extensions/skillsconf/src/ui/SkillConfView.ts +758 -758
  183. package/agent/extensions/skillsconf/src/ui/text-input.ts +91 -91
  184. package/agent/extensions/skillsconf/src/ui/tui-helpers.ts +144 -144
  185. package/agent/extensions/skillsconf/tsconfig.json +15 -15
  186. package/agent/extensions/statusline/index.ts +273 -282
  187. package/agent/extensions/statusline/install.ps1 +27 -27
  188. package/agent/extensions/statusline/install.sh +24 -24
  189. package/agent/extensions/subagents/by-the-way.test.ts +29 -29
  190. package/agent/extensions/subagents/claude.test.ts +119 -119
  191. package/agent/extensions/subagents/codex.test.ts +102 -102
  192. package/agent/extensions/subagents/context-usage.test.ts +107 -107
  193. package/agent/extensions/subagents/docs/design-plan.md +568 -568
  194. package/agent/extensions/subagents/docs/effect-v4-extension-guide.md +354 -354
  195. package/agent/extensions/subagents/docs/effect-v4-notes.md +571 -571
  196. package/agent/extensions/subagents/index.ts +779 -779
  197. package/agent/extensions/subagents/install.ps1 +27 -27
  198. package/agent/extensions/subagents/install.sh +24 -24
  199. package/agent/extensions/subagents/manager.test.ts +276 -276
  200. package/agent/extensions/subagents/package-lock.json +2244 -2244
  201. package/agent/extensions/subagents/package.json +19 -19
  202. package/agent/extensions/subagents/result-delivery.test.ts +27 -27
  203. package/agent/extensions/subagents/src/backend.ts +73 -73
  204. package/agent/extensions/subagents/src/backends/claude.ts +701 -701
  205. package/agent/extensions/subagents/src/backends/codex.ts +1060 -1060
  206. package/agent/extensions/subagents/src/backends/pi.ts +575 -575
  207. package/agent/extensions/subagents/src/backends/stub.ts +300 -300
  208. package/agent/extensions/subagents/src/by-the-way.ts +21 -21
  209. package/agent/extensions/subagents/src/domain.ts +253 -253
  210. package/agent/extensions/subagents/src/format.ts +74 -74
  211. package/agent/extensions/subagents/src/manager.ts +736 -736
  212. package/agent/extensions/subagents/src/prompt.ts +92 -92
  213. package/agent/extensions/subagents/src/result-delivery.ts +20 -20
  214. package/agent/extensions/subagents/src/runtime.ts +53 -53
  215. package/agent/extensions/subagents/src/ui/takeover.ts +583 -583
  216. package/agent/extensions/subagents/src/ui/transcript.ts +201 -201
  217. package/agent/extensions/subagents/takeover.test.ts +29 -29
  218. package/agent/extensions/subagents/tsconfig.json +7 -7
  219. package/agent/extensions/todo/AGENTS.md +38 -38
  220. package/agent/extensions/todo/LICENSE +21 -21
  221. package/agent/extensions/todo/config.ts +55 -55
  222. package/agent/extensions/todo/index.ts +151 -151
  223. package/agent/extensions/todo/install.ps1 +27 -27
  224. package/agent/extensions/todo/install.sh +24 -24
  225. package/agent/extensions/todo/locales/de.json +17 -17
  226. package/agent/extensions/todo/locales/en.json +15 -15
  227. package/agent/extensions/todo/locales/es.json +17 -17
  228. package/agent/extensions/todo/locales/fr.json +17 -17
  229. package/agent/extensions/todo/locales/pt-BR.json +17 -17
  230. package/agent/extensions/todo/locales/pt.json +17 -17
  231. package/agent/extensions/todo/locales/ru.json +17 -17
  232. package/agent/extensions/todo/locales/uk.json +17 -17
  233. package/agent/extensions/todo/locales/zh.json +17 -17
  234. package/agent/extensions/todo/package-lock.json +3358 -3358
  235. package/agent/extensions/todo/package.json +67 -67
  236. package/agent/extensions/todo/state/i18n-bridge.ts +64 -64
  237. package/agent/extensions/todo/state/invariants.ts +20 -20
  238. package/agent/extensions/todo/state/replay.ts +38 -38
  239. package/agent/extensions/todo/state/selectors.ts +107 -107
  240. package/agent/extensions/todo/state/state-reducer.ts +326 -326
  241. package/agent/extensions/todo/state/state.ts +18 -18
  242. package/agent/extensions/todo/state/store.ts +82 -82
  243. package/agent/extensions/todo/state/task-graph.ts +57 -57
  244. package/agent/extensions/todo/todo-overlay.ts +200 -200
  245. package/agent/extensions/todo/todo.ts +155 -155
  246. package/agent/extensions/todo/tool/response-envelope.ts +109 -109
  247. package/agent/extensions/todo/tool/types.ts +206 -206
  248. package/agent/extensions/todo/view/format.ts +177 -177
  249. package/agent/extensions/trim-context/README.md +54 -54
  250. package/agent/extensions/trim-context/index.ts +487 -487
  251. package/agent/extensions/trim-context/install.ps1 +27 -27
  252. package/agent/extensions/trim-context/install.sh +24 -24
  253. package/agent/keybindings.json +7 -7
  254. package/agent/version.txt +1 -1
  255. package/bin/kiln.js +20 -13
  256. package/package.json +1 -1
  257. package/agent/extensions/taste/index.ts +0 -443
  258. package/agent/extensions/taste/install.ps1 +0 -27
  259. package/agent/extensions/taste/install.sh +0 -24
@@ -1,552 +1,552 @@
1
- import assert from "node:assert/strict";
2
- import { after, test } from "node:test";
3
-
4
- const originalFetch = globalThis.fetch;
5
- const originalKey = process.env.DATALAB_API_KEY;
6
- const originalMode = process.env.DATALAB_MODE;
7
- const originalLocation = process.env.DATALAB_PROCESSING_LOCATION;
8
- const originalBase = process.env.DATALAB_API_BASE;
9
- process.env.DATALAB_API_KEY = "synthetic-datalab-key";
10
-
11
- const { extractPDFViaDatalab, getDatalabApiBase } = await import(
12
- "../datalab-pdf-extract.ts"
13
- );
14
-
15
- after(() => {
16
- globalThis.fetch = originalFetch;
17
- restoreEnv("DATALAB_API_KEY", originalKey);
18
- restoreEnv("DATALAB_MODE", originalMode);
19
- restoreEnv("DATALAB_PROCESSING_LOCATION", originalLocation);
20
- restoreEnv("DATALAB_API_BASE", originalBase);
21
- });
22
-
23
- test("Datalab conversion uploads the PDF, converts with a datalab reference, and polls to completion", async () => {
24
- const calls = [];
25
- globalThis.fetch = async (url, init) => {
26
- calls.push({ url: String(url), init });
27
- return route(url, init);
28
- };
29
-
30
- const input = Uint8Array.from([1, 2, 3, 4]).buffer;
31
- const result = await extractPDFViaDatalab(input, {
32
- maxPages: 5,
33
- title: "Sample Document",
34
- });
35
-
36
- const byUrl = calls.map((call) => call.url);
37
- assert.equal(byUrl.filter((url) => url.endsWith("/files/upload")).length, 1);
38
- assert.equal(byUrl.filter((url) => url.includes("/put/")).length, 1);
39
- assert.equal(
40
- byUrl.filter((url) => url.endsWith("/files/7/confirm")).length,
41
- 1,
42
- );
43
- assert.equal(byUrl.filter((url) => url.endsWith("/convert")).length, 1);
44
- assert.equal(byUrl.filter((url) => url.endsWith("/check/abc123")).length, 1);
45
-
46
- const upload = calls.find((call) => call.url.endsWith("/files/upload"));
47
- assert.equal(
48
- new Headers(upload.init.headers).get("x-api-key"),
49
- "synthetic-datalab-key",
50
- );
51
- assert.equal(
52
- new Headers(upload.init.headers).get("content-type"),
53
- "application/json",
54
- );
55
- const uploadBody = parseJsonBody(upload.init.body);
56
- assert.ok(uploadBody, "upload body did not parse");
57
- assert.equal(uploadBody.filename, "sample-document.pdf");
58
- assert.equal(uploadBody.content_type, "application/pdf");
59
- assert.equal(uploadBody.processing_location, "us");
60
-
61
- const put = calls.find((call) => call.url.includes("/put/"));
62
- assert.equal(put.init.method, "PUT");
63
- assert.equal(
64
- new Headers(put.init.headers).get("content-type"),
65
- "application/pdf",
66
- );
67
-
68
- const convert = calls.find((call) => call.url.endsWith("/convert"));
69
- assert.equal(convert.init.method, "POST");
70
- assert.equal(
71
- new Headers(convert.init.headers).get("x-api-key"),
72
- "synthetic-datalab-key",
73
- );
74
- const body = convert.init.body;
75
- assert.ok(body instanceof FormData);
76
- assert.equal(body.get("file_url"), "datalab://file-7");
77
- assert.equal(body.get("output_format"), "markdown");
78
- assert.equal(body.get("paginate"), "true");
79
- assert.equal(body.get("max_pages"), "5");
80
- assert.equal(body.get("mode"), "balanced");
81
- assert.equal(body.get("processing_location"), null);
82
-
83
- const cleanup = calls.find(
84
- (call) => call.url.endsWith("/files/7") && call.init.method === "DELETE",
85
- );
86
- assert.ok(cleanup, "expected best-effort file cleanup after conversion");
87
-
88
- assert.match(result.markdown, /^<!-- Page 1 -->/);
89
- assert.equal(result.pages, 2);
90
- assert.equal(result.parseQualityScore, 4.5);
91
- });
92
-
93
- test("Datalab conversion still succeeds when cleanup fails", async () => {
94
- globalThis.fetch = async (url, init) => {
95
- if (String(url).endsWith("/files/7") && init.method === "DELETE") {
96
- throw new Error("cleanup transport failure");
97
- }
98
- return route(url, init);
99
- };
100
-
101
- const result = await extractPDFViaDatalab(new ArrayBuffer(1), {
102
- maxPages: 1,
103
- title: "Doc",
104
- });
105
- assert.match(result.markdown, /Sample Document/);
106
- });
107
-
108
- test("Datalab conversion honors processing location and mode selection", async () => {
109
- let capturedLocation = null;
110
- let capturedMode = null;
111
- globalThis.fetch = async (url, init) => {
112
- if (String(url).endsWith("/files/upload")) {
113
- capturedLocation = parseJsonBody(init.body)?.processing_location ?? null;
114
- return jsonResponse({
115
- file_id: 7,
116
- upload_url: "https://storage.test/put/abc",
117
- reference: "datalab://file-7",
118
- });
119
- }
120
- if (String(url).endsWith("/convert")) {
121
- const body = init.body;
122
- capturedMode = body.get("mode");
123
- return jsonResponse({
124
- status: "complete",
125
- success: true,
126
- markdown: "<!-- Page 1 -->\nDone",
127
- page_count: 1,
128
- });
129
- }
130
- return route(url, init);
131
- };
132
-
133
- process.env.DATALAB_PROCESSING_LOCATION = "eu";
134
- process.env.DATALAB_MODE = "accurate";
135
- await extractPDFViaDatalab(new ArrayBuffer(1), { maxPages: 2, title: "Doc" });
136
-
137
- assert.equal(capturedLocation, "eu");
138
- assert.equal(capturedMode, "accurate");
139
- delete process.env.DATALAB_PROCESSING_LOCATION;
140
- delete process.env.DATALAB_MODE;
141
- });
142
-
143
- test("Datalab conversion rejects a missing file_id before uploading", async () => {
144
- const calls = [];
145
- globalThis.fetch = async (url, init) => {
146
- calls.push(String(url));
147
- if (String(url).endsWith("/files/upload")) {
148
- return jsonResponse({
149
- upload_url: "https://storage.test/put/abc",
150
- reference: "datalab://file-7",
151
- });
152
- }
153
- return route(url, init);
154
- };
155
-
156
- await assert.rejects(
157
- extractPDFViaDatalab(new ArrayBuffer(1), { maxPages: 1, title: "Doc" }),
158
- /missing file_id/,
159
- );
160
- assert.equal(
161
- calls.some((url) => url.includes("/put/")),
162
- false,
163
- "must not upload without a file ID for confirmation and cleanup",
164
- );
165
- });
166
-
167
- test("Datalab conversion rejects valid JSON that is not an object", async () => {
168
- globalThis.fetch = async (url, init) => {
169
- if (String(url).endsWith("/files/upload")) {
170
- return new Response("null", {
171
- status: 200,
172
- headers: { "content-type": "application/json" },
173
- });
174
- }
175
- return route(url, init);
176
- };
177
-
178
- await assert.rejects(
179
- extractPDFViaDatalab(new ArrayBuffer(1), { maxPages: 1, title: "Doc" }),
180
- /invalid JSON object/,
181
- );
182
- });
183
-
184
- test("Datalab conversion returns immediately when the convert response is complete", async () => {
185
- let convertCalls = 0;
186
- globalThis.fetch = async (url, init) => {
187
- if (String(url).endsWith("/convert")) {
188
- convertCalls += 1;
189
- return jsonResponse({
190
- status: "complete",
191
- success: true,
192
- markdown: "<!-- Page 1 -->\nFast",
193
- page_count: 1,
194
- });
195
- }
196
- return route(url, init);
197
- };
198
-
199
- const result = await extractPDFViaDatalab(new ArrayBuffer(1), {
200
- maxPages: 1,
201
- title: "Doc",
202
- });
203
- assert.equal(convertCalls, 1);
204
- assert.match(result.markdown, /Fast/);
205
- });
206
-
207
- test("Datalab conversion rejects HTTP errors with the status and redacted body", async () => {
208
- globalThis.fetch = async (url, init) => {
209
- if (String(url).endsWith("/files/upload")) {
210
- return new Response(JSON.stringify({ detail: "invalid api key" }), {
211
- status: 401,
212
- statusText: "Unauthorized",
213
- headers: { "content-type": "application/json" },
214
- });
215
- }
216
- return route(url, init);
217
- };
218
-
219
- await assert.rejects(
220
- extractPDFViaDatalab(new ArrayBuffer(1), { maxPages: 1, title: "Doc" }),
221
- /HTTP 401 Unauthorized: .*invalid api key/,
222
- );
223
- });
224
-
225
- test("Datalab conversion redacts API keys from transport errors", async () => {
226
- globalThis.fetch = async () => {
227
- throw new Error("network failure for synthetic-datalab-key");
228
- };
229
-
230
- await assert.rejects(
231
- extractPDFViaDatalab(new ArrayBuffer(1), { maxPages: 1, title: "Doc" }),
232
- (error) => {
233
- assert.equal(error.name, "Error");
234
- assert.doesNotMatch(error.message, /synthetic-datalab-key/);
235
- return true;
236
- },
237
- );
238
- });
239
-
240
- test("Datalab conversion rejects a streamed response above its byte limit", async () => {
241
- globalThis.fetch = async (url, init) => {
242
- if (String(url).endsWith("/files/upload")) {
243
- return new Response(
244
- new ReadableStream({
245
- start(controller) {
246
- controller.enqueue(new Uint8Array(4 * 1024 * 1024 + 1));
247
- controller.close();
248
- },
249
- }),
250
- { status: 200 },
251
- );
252
- }
253
- return route(url, init);
254
- };
255
-
256
- await assert.rejects(
257
- extractPDFViaDatalab(new ArrayBuffer(1), { maxPages: 1, title: "Doc" }),
258
- /response too large/,
259
- );
260
- });
261
-
262
- test("Datalab conversion rejects failed conversion status", async () => {
263
- globalThis.fetch = async (url, init) => {
264
- if (String(url).endsWith("/convert")) {
265
- return jsonResponse({
266
- status: "processing",
267
- request_check_url: "/check/abc",
268
- });
269
- }
270
- if (String(url).endsWith("/check/abc")) {
271
- return jsonResponse({
272
- status: "failed",
273
- success: false,
274
- error: "document is corrupt",
275
- });
276
- }
277
- return route(url, init);
278
- };
279
-
280
- await assert.rejects(
281
- extractPDFViaDatalab(new ArrayBuffer(1), { maxPages: 1, title: "Doc" }),
282
- /document is corrupt/,
283
- );
284
- });
285
-
286
- test("Datalab conversion rejects empty markdown", async () => {
287
- globalThis.fetch = async (url, init) => {
288
- if (String(url).endsWith("/convert")) {
289
- return jsonResponse({
290
- status: "complete",
291
- success: true,
292
- markdown: " ",
293
- page_count: 1,
294
- });
295
- }
296
- return route(url, init);
297
- };
298
-
299
- await assert.rejects(
300
- extractPDFViaDatalab(new ArrayBuffer(1), { maxPages: 1, title: "Doc" }),
301
- /empty markdown/,
302
- );
303
- });
304
-
305
- test("Datalab conversion times out while polling without waiting a full poll interval", async () => {
306
- globalThis.fetch = async (url, init) => {
307
- if (String(url).endsWith("/convert")) {
308
- return jsonResponse({
309
- status: "processing",
310
- request_check_url: "/check/abc",
311
- });
312
- }
313
- if (String(url).endsWith("/check/abc")) {
314
- return jsonResponse({ status: "processing" });
315
- }
316
- return route(url, init);
317
- };
318
-
319
- const started = Date.now();
320
- await assert.rejects(
321
- extractPDFViaDatalab(new ArrayBuffer(1), {
322
- maxPages: 1,
323
- title: "Doc",
324
- timeoutMs: 60,
325
- }),
326
- /timed out/,
327
- );
328
- assert.ok(
329
- Date.now() - started < 500,
330
- "polling must not delay a 60ms timeout by its 1.5s interval",
331
- );
332
- });
333
-
334
- test("Datalab cancellation returns without awaiting best-effort cleanup", async () => {
335
- const controller = new AbortController();
336
- let markConvertSubmitted;
337
- const convertSubmitted = new Promise((resolve) => {
338
- markConvertSubmitted = resolve;
339
- });
340
- let markCleanupStarted;
341
- const cleanupStarted = new Promise((resolve) => {
342
- markCleanupStarted = resolve;
343
- });
344
- let finishCleanup;
345
- const cleanupBlocked = new Promise((resolve) => {
346
- finishCleanup = resolve;
347
- });
348
-
349
- globalThis.fetch = async (url, init) => {
350
- if (String(url).endsWith("/convert")) markConvertSubmitted();
351
- if (String(url).endsWith("/files/7") && init.method === "DELETE") {
352
- markCleanupStarted();
353
- return cleanupBlocked;
354
- }
355
- return route(url, init);
356
- };
357
-
358
- const extraction = extractPDFViaDatalab(new ArrayBuffer(1), {
359
- maxPages: 1,
360
- title: "Doc",
361
- signal: controller.signal,
362
- });
363
- await convertSubmitted;
364
- controller.abort(new DOMException("cancelled", "AbortError"));
365
- await cleanupStarted;
366
-
367
- const outcome = await settlesWithin(extraction, 100);
368
- assert.equal(outcome.status, "rejected");
369
- assert.equal(outcome.error.name, "AbortError");
370
- finishCleanup(new Response(null, { status: 200 }));
371
- });
372
-
373
- test("Datalab conversion rejects a cross-origin polling URL", async () => {
374
- const calls = [];
375
- globalThis.fetch = async (url, init) => {
376
- calls.push(String(url));
377
- if (String(url).endsWith("/convert")) {
378
- return jsonResponse({
379
- status: "processing",
380
- request_check_url: "https://untrusted.example.test/check/abc",
381
- });
382
- }
383
- return route(url, init);
384
- };
385
-
386
- await assert.rejects(
387
- extractPDFViaDatalab(new ArrayBuffer(1), { maxPages: 1, title: "Doc" }),
388
- /unexpected origin/,
389
- );
390
- assert.equal(
391
- calls.some((url) => url.includes("untrusted.example.test")),
392
- false,
393
- "must not send the API key to a polling URL returned by another origin",
394
- );
395
- });
396
-
397
- test("Datalab conversion surfaces missing request_check_url", async () => {
398
- globalThis.fetch = async (url, init) => {
399
- if (String(url).endsWith("/convert")) {
400
- return jsonResponse({ status: "processing" });
401
- }
402
- return route(url, init);
403
- };
404
-
405
- await assert.rejects(
406
- extractPDFViaDatalab(new ArrayBuffer(1), { maxPages: 1, title: "Doc" }),
407
- /missing request_check_url/,
408
- );
409
- });
410
-
411
- test("Datalab conversion honors a custom API base", async () => {
412
- process.env.DATALAB_API_BASE = "https://datalab.example.test/api/v1";
413
- const seen = [];
414
- globalThis.fetch = async (url, init) => {
415
- seen.push(String(url));
416
- if (String(url).endsWith("/check/abc123")) {
417
- return jsonResponse({
418
- status: "complete",
419
- success: true,
420
- markdown: "<!-- Page 1 -->\nCustom",
421
- page_count: 1,
422
- });
423
- }
424
- return route(url, init);
425
- };
426
-
427
- assert.equal(getDatalabApiBase(), "https://datalab.example.test/api/v1");
428
- const result = await extractPDFViaDatalab(new ArrayBuffer(1), {
429
- maxPages: 1,
430
- title: "Doc",
431
- });
432
- assert.match(result.markdown, /Custom/);
433
- assert.ok(
434
- seen.some((url) =>
435
- url.startsWith("https://datalab.example.test/api/v1/files/upload"),
436
- ),
437
- );
438
- assert.ok(
439
- seen.some((url) =>
440
- url.startsWith("https://datalab.example.test/api/v1/convert"),
441
- ),
442
- );
443
- delete process.env.DATALAB_API_BASE;
444
- });
445
-
446
- test("Datalab conversion preserves caller cancellation", async () => {
447
- const controller = new AbortController();
448
- controller.abort(new DOMException("cancelled", "AbortError"));
449
-
450
- await assert.rejects(
451
- extractPDFViaDatalab(new ArrayBuffer(1), {
452
- maxPages: 1,
453
- title: "Doc",
454
- signal: controller.signal,
455
- }),
456
- (error) => error instanceof DOMException && error.name === "AbortError",
457
- );
458
- });
459
-
460
- test("Datalab conversion rejects invalid processing location env", async () => {
461
- process.env.DATALAB_PROCESSING_LOCATION = "mars";
462
- await assert.rejects(
463
- extractPDFViaDatalab(new ArrayBuffer(1), { maxPages: 1, title: "Doc" }),
464
- /Failed to parse DATALAB_PROCESSING_LOCATION/,
465
- );
466
- delete process.env.DATALAB_PROCESSING_LOCATION;
467
- });
468
-
469
- function route(url) {
470
- const urlString = String(url);
471
- if (urlString.endsWith("/files/upload")) {
472
- return jsonResponse({
473
- file_id: 7,
474
- upload_url: "https://storage.test/put/abc",
475
- reference: "datalab://file-7",
476
- expires_in: 3600,
477
- });
478
- }
479
- if (urlString.includes("/put/")) {
480
- return new Response(null, { status: 200 });
481
- }
482
- if (urlString.endsWith("/files/7/confirm")) {
483
- return jsonResponse({
484
- file_id: 7,
485
- reference: "datalab://file-7",
486
- message: "confirmed",
487
- });
488
- }
489
- if (urlString.endsWith("/files/7")) {
490
- return new Response(null, { status: 200 });
491
- }
492
- if (urlString.endsWith("/convert")) {
493
- return jsonResponse({
494
- status: "processing",
495
- request_check_url: "/check/abc123",
496
- });
497
- }
498
- if (urlString.endsWith("/check/abc123")) {
499
- return jsonResponse({
500
- status: "complete",
501
- success: true,
502
- markdown:
503
- "<!-- Page 1 -->\n# Sample Document\n\nFirst page.\n\n<!-- Page 2 -->\n\nSecond page.",
504
- page_count: 2,
505
- parse_quality_score: 4.5,
506
- cost_breakdown: { total_cost_cents: 1 },
507
- });
508
- }
509
- return new Response(
510
- JSON.stringify({ detail: "unexpected fetch: " + urlString }),
511
- {
512
- status: 418,
513
- headers: { "content-type": "application/json" },
514
- },
515
- );
516
- }
517
-
518
- function parseJsonBody(value) {
519
- try {
520
- return JSON.parse(value);
521
- } catch {
522
- return null;
523
- }
524
- }
525
-
526
- function jsonResponse(value) {
527
- return new Response(JSON.stringify(value), {
528
- status: 200,
529
- headers: { "content-type": "application/json" },
530
- });
531
- }
532
-
533
- function settlesWithin(promise, timeoutMs) {
534
- return new Promise((resolve) => {
535
- const timeout = setTimeout(() => resolve({ status: "pending" }), timeoutMs);
536
- promise.then(
537
- () => {
538
- clearTimeout(timeout);
539
- resolve({ status: "resolved" });
540
- },
541
- (error) => {
542
- clearTimeout(timeout);
543
- resolve({ status: "rejected", error });
544
- },
545
- );
546
- });
547
- }
548
-
549
- function restoreEnv(name, value) {
550
- if (value === undefined) delete process.env[name];
551
- else process.env[name] = value;
552
- }
1
+ import assert from "node:assert/strict";
2
+ import { after, test } from "node:test";
3
+
4
+ const originalFetch = globalThis.fetch;
5
+ const originalKey = process.env.DATALAB_API_KEY;
6
+ const originalMode = process.env.DATALAB_MODE;
7
+ const originalLocation = process.env.DATALAB_PROCESSING_LOCATION;
8
+ const originalBase = process.env.DATALAB_API_BASE;
9
+ process.env.DATALAB_API_KEY = "synthetic-datalab-key";
10
+
11
+ const { extractPDFViaDatalab, getDatalabApiBase } = await import(
12
+ "../datalab-pdf-extract.ts"
13
+ );
14
+
15
+ after(() => {
16
+ globalThis.fetch = originalFetch;
17
+ restoreEnv("DATALAB_API_KEY", originalKey);
18
+ restoreEnv("DATALAB_MODE", originalMode);
19
+ restoreEnv("DATALAB_PROCESSING_LOCATION", originalLocation);
20
+ restoreEnv("DATALAB_API_BASE", originalBase);
21
+ });
22
+
23
+ test("Datalab conversion uploads the PDF, converts with a datalab reference, and polls to completion", async () => {
24
+ const calls = [];
25
+ globalThis.fetch = async (url, init) => {
26
+ calls.push({ url: String(url), init });
27
+ return route(url, init);
28
+ };
29
+
30
+ const input = Uint8Array.from([1, 2, 3, 4]).buffer;
31
+ const result = await extractPDFViaDatalab(input, {
32
+ maxPages: 5,
33
+ title: "Sample Document",
34
+ });
35
+
36
+ const byUrl = calls.map((call) => call.url);
37
+ assert.equal(byUrl.filter((url) => url.endsWith("/files/upload")).length, 1);
38
+ assert.equal(byUrl.filter((url) => url.includes("/put/")).length, 1);
39
+ assert.equal(
40
+ byUrl.filter((url) => url.endsWith("/files/7/confirm")).length,
41
+ 1,
42
+ );
43
+ assert.equal(byUrl.filter((url) => url.endsWith("/convert")).length, 1);
44
+ assert.equal(byUrl.filter((url) => url.endsWith("/check/abc123")).length, 1);
45
+
46
+ const upload = calls.find((call) => call.url.endsWith("/files/upload"));
47
+ assert.equal(
48
+ new Headers(upload.init.headers).get("x-api-key"),
49
+ "synthetic-datalab-key",
50
+ );
51
+ assert.equal(
52
+ new Headers(upload.init.headers).get("content-type"),
53
+ "application/json",
54
+ );
55
+ const uploadBody = parseJsonBody(upload.init.body);
56
+ assert.ok(uploadBody, "upload body did not parse");
57
+ assert.equal(uploadBody.filename, "sample-document.pdf");
58
+ assert.equal(uploadBody.content_type, "application/pdf");
59
+ assert.equal(uploadBody.processing_location, "us");
60
+
61
+ const put = calls.find((call) => call.url.includes("/put/"));
62
+ assert.equal(put.init.method, "PUT");
63
+ assert.equal(
64
+ new Headers(put.init.headers).get("content-type"),
65
+ "application/pdf",
66
+ );
67
+
68
+ const convert = calls.find((call) => call.url.endsWith("/convert"));
69
+ assert.equal(convert.init.method, "POST");
70
+ assert.equal(
71
+ new Headers(convert.init.headers).get("x-api-key"),
72
+ "synthetic-datalab-key",
73
+ );
74
+ const body = convert.init.body;
75
+ assert.ok(body instanceof FormData);
76
+ assert.equal(body.get("file_url"), "datalab://file-7");
77
+ assert.equal(body.get("output_format"), "markdown");
78
+ assert.equal(body.get("paginate"), "true");
79
+ assert.equal(body.get("max_pages"), "5");
80
+ assert.equal(body.get("mode"), "balanced");
81
+ assert.equal(body.get("processing_location"), null);
82
+
83
+ const cleanup = calls.find(
84
+ (call) => call.url.endsWith("/files/7") && call.init.method === "DELETE",
85
+ );
86
+ assert.ok(cleanup, "expected best-effort file cleanup after conversion");
87
+
88
+ assert.match(result.markdown, /^<!-- Page 1 -->/);
89
+ assert.equal(result.pages, 2);
90
+ assert.equal(result.parseQualityScore, 4.5);
91
+ });
92
+
93
+ test("Datalab conversion still succeeds when cleanup fails", async () => {
94
+ globalThis.fetch = async (url, init) => {
95
+ if (String(url).endsWith("/files/7") && init.method === "DELETE") {
96
+ throw new Error("cleanup transport failure");
97
+ }
98
+ return route(url, init);
99
+ };
100
+
101
+ const result = await extractPDFViaDatalab(new ArrayBuffer(1), {
102
+ maxPages: 1,
103
+ title: "Doc",
104
+ });
105
+ assert.match(result.markdown, /Sample Document/);
106
+ });
107
+
108
+ test("Datalab conversion honors processing location and mode selection", async () => {
109
+ let capturedLocation = null;
110
+ let capturedMode = null;
111
+ globalThis.fetch = async (url, init) => {
112
+ if (String(url).endsWith("/files/upload")) {
113
+ capturedLocation = parseJsonBody(init.body)?.processing_location ?? null;
114
+ return jsonResponse({
115
+ file_id: 7,
116
+ upload_url: "https://storage.test/put/abc",
117
+ reference: "datalab://file-7",
118
+ });
119
+ }
120
+ if (String(url).endsWith("/convert")) {
121
+ const body = init.body;
122
+ capturedMode = body.get("mode");
123
+ return jsonResponse({
124
+ status: "complete",
125
+ success: true,
126
+ markdown: "<!-- Page 1 -->\nDone",
127
+ page_count: 1,
128
+ });
129
+ }
130
+ return route(url, init);
131
+ };
132
+
133
+ process.env.DATALAB_PROCESSING_LOCATION = "eu";
134
+ process.env.DATALAB_MODE = "accurate";
135
+ await extractPDFViaDatalab(new ArrayBuffer(1), { maxPages: 2, title: "Doc" });
136
+
137
+ assert.equal(capturedLocation, "eu");
138
+ assert.equal(capturedMode, "accurate");
139
+ delete process.env.DATALAB_PROCESSING_LOCATION;
140
+ delete process.env.DATALAB_MODE;
141
+ });
142
+
143
+ test("Datalab conversion rejects a missing file_id before uploading", async () => {
144
+ const calls = [];
145
+ globalThis.fetch = async (url, init) => {
146
+ calls.push(String(url));
147
+ if (String(url).endsWith("/files/upload")) {
148
+ return jsonResponse({
149
+ upload_url: "https://storage.test/put/abc",
150
+ reference: "datalab://file-7",
151
+ });
152
+ }
153
+ return route(url, init);
154
+ };
155
+
156
+ await assert.rejects(
157
+ extractPDFViaDatalab(new ArrayBuffer(1), { maxPages: 1, title: "Doc" }),
158
+ /missing file_id/,
159
+ );
160
+ assert.equal(
161
+ calls.some((url) => url.includes("/put/")),
162
+ false,
163
+ "must not upload without a file ID for confirmation and cleanup",
164
+ );
165
+ });
166
+
167
+ test("Datalab conversion rejects valid JSON that is not an object", async () => {
168
+ globalThis.fetch = async (url, init) => {
169
+ if (String(url).endsWith("/files/upload")) {
170
+ return new Response("null", {
171
+ status: 200,
172
+ headers: { "content-type": "application/json" },
173
+ });
174
+ }
175
+ return route(url, init);
176
+ };
177
+
178
+ await assert.rejects(
179
+ extractPDFViaDatalab(new ArrayBuffer(1), { maxPages: 1, title: "Doc" }),
180
+ /invalid JSON object/,
181
+ );
182
+ });
183
+
184
+ test("Datalab conversion returns immediately when the convert response is complete", async () => {
185
+ let convertCalls = 0;
186
+ globalThis.fetch = async (url, init) => {
187
+ if (String(url).endsWith("/convert")) {
188
+ convertCalls += 1;
189
+ return jsonResponse({
190
+ status: "complete",
191
+ success: true,
192
+ markdown: "<!-- Page 1 -->\nFast",
193
+ page_count: 1,
194
+ });
195
+ }
196
+ return route(url, init);
197
+ };
198
+
199
+ const result = await extractPDFViaDatalab(new ArrayBuffer(1), {
200
+ maxPages: 1,
201
+ title: "Doc",
202
+ });
203
+ assert.equal(convertCalls, 1);
204
+ assert.match(result.markdown, /Fast/);
205
+ });
206
+
207
+ test("Datalab conversion rejects HTTP errors with the status and redacted body", async () => {
208
+ globalThis.fetch = async (url, init) => {
209
+ if (String(url).endsWith("/files/upload")) {
210
+ return new Response(JSON.stringify({ detail: "invalid api key" }), {
211
+ status: 401,
212
+ statusText: "Unauthorized",
213
+ headers: { "content-type": "application/json" },
214
+ });
215
+ }
216
+ return route(url, init);
217
+ };
218
+
219
+ await assert.rejects(
220
+ extractPDFViaDatalab(new ArrayBuffer(1), { maxPages: 1, title: "Doc" }),
221
+ /HTTP 401 Unauthorized: .*invalid api key/,
222
+ );
223
+ });
224
+
225
+ test("Datalab conversion redacts API keys from transport errors", async () => {
226
+ globalThis.fetch = async () => {
227
+ throw new Error("network failure for synthetic-datalab-key");
228
+ };
229
+
230
+ await assert.rejects(
231
+ extractPDFViaDatalab(new ArrayBuffer(1), { maxPages: 1, title: "Doc" }),
232
+ (error) => {
233
+ assert.equal(error.name, "Error");
234
+ assert.doesNotMatch(error.message, /synthetic-datalab-key/);
235
+ return true;
236
+ },
237
+ );
238
+ });
239
+
240
+ test("Datalab conversion rejects a streamed response above its byte limit", async () => {
241
+ globalThis.fetch = async (url, init) => {
242
+ if (String(url).endsWith("/files/upload")) {
243
+ return new Response(
244
+ new ReadableStream({
245
+ start(controller) {
246
+ controller.enqueue(new Uint8Array(4 * 1024 * 1024 + 1));
247
+ controller.close();
248
+ },
249
+ }),
250
+ { status: 200 },
251
+ );
252
+ }
253
+ return route(url, init);
254
+ };
255
+
256
+ await assert.rejects(
257
+ extractPDFViaDatalab(new ArrayBuffer(1), { maxPages: 1, title: "Doc" }),
258
+ /response too large/,
259
+ );
260
+ });
261
+
262
+ test("Datalab conversion rejects failed conversion status", async () => {
263
+ globalThis.fetch = async (url, init) => {
264
+ if (String(url).endsWith("/convert")) {
265
+ return jsonResponse({
266
+ status: "processing",
267
+ request_check_url: "/check/abc",
268
+ });
269
+ }
270
+ if (String(url).endsWith("/check/abc")) {
271
+ return jsonResponse({
272
+ status: "failed",
273
+ success: false,
274
+ error: "document is corrupt",
275
+ });
276
+ }
277
+ return route(url, init);
278
+ };
279
+
280
+ await assert.rejects(
281
+ extractPDFViaDatalab(new ArrayBuffer(1), { maxPages: 1, title: "Doc" }),
282
+ /document is corrupt/,
283
+ );
284
+ });
285
+
286
+ test("Datalab conversion rejects empty markdown", async () => {
287
+ globalThis.fetch = async (url, init) => {
288
+ if (String(url).endsWith("/convert")) {
289
+ return jsonResponse({
290
+ status: "complete",
291
+ success: true,
292
+ markdown: " ",
293
+ page_count: 1,
294
+ });
295
+ }
296
+ return route(url, init);
297
+ };
298
+
299
+ await assert.rejects(
300
+ extractPDFViaDatalab(new ArrayBuffer(1), { maxPages: 1, title: "Doc" }),
301
+ /empty markdown/,
302
+ );
303
+ });
304
+
305
+ test("Datalab conversion times out while polling without waiting a full poll interval", async () => {
306
+ globalThis.fetch = async (url, init) => {
307
+ if (String(url).endsWith("/convert")) {
308
+ return jsonResponse({
309
+ status: "processing",
310
+ request_check_url: "/check/abc",
311
+ });
312
+ }
313
+ if (String(url).endsWith("/check/abc")) {
314
+ return jsonResponse({ status: "processing" });
315
+ }
316
+ return route(url, init);
317
+ };
318
+
319
+ const started = Date.now();
320
+ await assert.rejects(
321
+ extractPDFViaDatalab(new ArrayBuffer(1), {
322
+ maxPages: 1,
323
+ title: "Doc",
324
+ timeoutMs: 60,
325
+ }),
326
+ /timed out/,
327
+ );
328
+ assert.ok(
329
+ Date.now() - started < 500,
330
+ "polling must not delay a 60ms timeout by its 1.5s interval",
331
+ );
332
+ });
333
+
334
+ test("Datalab cancellation returns without awaiting best-effort cleanup", async () => {
335
+ const controller = new AbortController();
336
+ let markConvertSubmitted;
337
+ const convertSubmitted = new Promise((resolve) => {
338
+ markConvertSubmitted = resolve;
339
+ });
340
+ let markCleanupStarted;
341
+ const cleanupStarted = new Promise((resolve) => {
342
+ markCleanupStarted = resolve;
343
+ });
344
+ let finishCleanup;
345
+ const cleanupBlocked = new Promise((resolve) => {
346
+ finishCleanup = resolve;
347
+ });
348
+
349
+ globalThis.fetch = async (url, init) => {
350
+ if (String(url).endsWith("/convert")) markConvertSubmitted();
351
+ if (String(url).endsWith("/files/7") && init.method === "DELETE") {
352
+ markCleanupStarted();
353
+ return cleanupBlocked;
354
+ }
355
+ return route(url, init);
356
+ };
357
+
358
+ const extraction = extractPDFViaDatalab(new ArrayBuffer(1), {
359
+ maxPages: 1,
360
+ title: "Doc",
361
+ signal: controller.signal,
362
+ });
363
+ await convertSubmitted;
364
+ controller.abort(new DOMException("cancelled", "AbortError"));
365
+ await cleanupStarted;
366
+
367
+ const outcome = await settlesWithin(extraction, 100);
368
+ assert.equal(outcome.status, "rejected");
369
+ assert.equal(outcome.error.name, "AbortError");
370
+ finishCleanup(new Response(null, { status: 200 }));
371
+ });
372
+
373
+ test("Datalab conversion rejects a cross-origin polling URL", async () => {
374
+ const calls = [];
375
+ globalThis.fetch = async (url, init) => {
376
+ calls.push(String(url));
377
+ if (String(url).endsWith("/convert")) {
378
+ return jsonResponse({
379
+ status: "processing",
380
+ request_check_url: "https://untrusted.example.test/check/abc",
381
+ });
382
+ }
383
+ return route(url, init);
384
+ };
385
+
386
+ await assert.rejects(
387
+ extractPDFViaDatalab(new ArrayBuffer(1), { maxPages: 1, title: "Doc" }),
388
+ /unexpected origin/,
389
+ );
390
+ assert.equal(
391
+ calls.some((url) => url.includes("untrusted.example.test")),
392
+ false,
393
+ "must not send the API key to a polling URL returned by another origin",
394
+ );
395
+ });
396
+
397
+ test("Datalab conversion surfaces missing request_check_url", async () => {
398
+ globalThis.fetch = async (url, init) => {
399
+ if (String(url).endsWith("/convert")) {
400
+ return jsonResponse({ status: "processing" });
401
+ }
402
+ return route(url, init);
403
+ };
404
+
405
+ await assert.rejects(
406
+ extractPDFViaDatalab(new ArrayBuffer(1), { maxPages: 1, title: "Doc" }),
407
+ /missing request_check_url/,
408
+ );
409
+ });
410
+
411
+ test("Datalab conversion honors a custom API base", async () => {
412
+ process.env.DATALAB_API_BASE = "https://datalab.example.test/api/v1";
413
+ const seen = [];
414
+ globalThis.fetch = async (url, init) => {
415
+ seen.push(String(url));
416
+ if (String(url).endsWith("/check/abc123")) {
417
+ return jsonResponse({
418
+ status: "complete",
419
+ success: true,
420
+ markdown: "<!-- Page 1 -->\nCustom",
421
+ page_count: 1,
422
+ });
423
+ }
424
+ return route(url, init);
425
+ };
426
+
427
+ assert.equal(getDatalabApiBase(), "https://datalab.example.test/api/v1");
428
+ const result = await extractPDFViaDatalab(new ArrayBuffer(1), {
429
+ maxPages: 1,
430
+ title: "Doc",
431
+ });
432
+ assert.match(result.markdown, /Custom/);
433
+ assert.ok(
434
+ seen.some((url) =>
435
+ url.startsWith("https://datalab.example.test/api/v1/files/upload"),
436
+ ),
437
+ );
438
+ assert.ok(
439
+ seen.some((url) =>
440
+ url.startsWith("https://datalab.example.test/api/v1/convert"),
441
+ ),
442
+ );
443
+ delete process.env.DATALAB_API_BASE;
444
+ });
445
+
446
+ test("Datalab conversion preserves caller cancellation", async () => {
447
+ const controller = new AbortController();
448
+ controller.abort(new DOMException("cancelled", "AbortError"));
449
+
450
+ await assert.rejects(
451
+ extractPDFViaDatalab(new ArrayBuffer(1), {
452
+ maxPages: 1,
453
+ title: "Doc",
454
+ signal: controller.signal,
455
+ }),
456
+ (error) => error instanceof DOMException && error.name === "AbortError",
457
+ );
458
+ });
459
+
460
+ test("Datalab conversion rejects invalid processing location env", async () => {
461
+ process.env.DATALAB_PROCESSING_LOCATION = "mars";
462
+ await assert.rejects(
463
+ extractPDFViaDatalab(new ArrayBuffer(1), { maxPages: 1, title: "Doc" }),
464
+ /Failed to parse DATALAB_PROCESSING_LOCATION/,
465
+ );
466
+ delete process.env.DATALAB_PROCESSING_LOCATION;
467
+ });
468
+
469
+ function route(url) {
470
+ const urlString = String(url);
471
+ if (urlString.endsWith("/files/upload")) {
472
+ return jsonResponse({
473
+ file_id: 7,
474
+ upload_url: "https://storage.test/put/abc",
475
+ reference: "datalab://file-7",
476
+ expires_in: 3600,
477
+ });
478
+ }
479
+ if (urlString.includes("/put/")) {
480
+ return new Response(null, { status: 200 });
481
+ }
482
+ if (urlString.endsWith("/files/7/confirm")) {
483
+ return jsonResponse({
484
+ file_id: 7,
485
+ reference: "datalab://file-7",
486
+ message: "confirmed",
487
+ });
488
+ }
489
+ if (urlString.endsWith("/files/7")) {
490
+ return new Response(null, { status: 200 });
491
+ }
492
+ if (urlString.endsWith("/convert")) {
493
+ return jsonResponse({
494
+ status: "processing",
495
+ request_check_url: "/check/abc123",
496
+ });
497
+ }
498
+ if (urlString.endsWith("/check/abc123")) {
499
+ return jsonResponse({
500
+ status: "complete",
501
+ success: true,
502
+ markdown:
503
+ "<!-- Page 1 -->\n# Sample Document\n\nFirst page.\n\n<!-- Page 2 -->\n\nSecond page.",
504
+ page_count: 2,
505
+ parse_quality_score: 4.5,
506
+ cost_breakdown: { total_cost_cents: 1 },
507
+ });
508
+ }
509
+ return new Response(
510
+ JSON.stringify({ detail: "unexpected fetch: " + urlString }),
511
+ {
512
+ status: 418,
513
+ headers: { "content-type": "application/json" },
514
+ },
515
+ );
516
+ }
517
+
518
+ function parseJsonBody(value) {
519
+ try {
520
+ return JSON.parse(value);
521
+ } catch {
522
+ return null;
523
+ }
524
+ }
525
+
526
+ function jsonResponse(value) {
527
+ return new Response(JSON.stringify(value), {
528
+ status: 200,
529
+ headers: { "content-type": "application/json" },
530
+ });
531
+ }
532
+
533
+ function settlesWithin(promise, timeoutMs) {
534
+ return new Promise((resolve) => {
535
+ const timeout = setTimeout(() => resolve({ status: "pending" }), timeoutMs);
536
+ promise.then(
537
+ () => {
538
+ clearTimeout(timeout);
539
+ resolve({ status: "resolved" });
540
+ },
541
+ (error) => {
542
+ clearTimeout(timeout);
543
+ resolve({ status: "rejected", error });
544
+ },
545
+ );
546
+ });
547
+ }
548
+
549
+ function restoreEnv(name, value) {
550
+ if (value === undefined) delete process.env[name];
551
+ else process.env[name] = value;
552
+ }