@modusensus/dsh-mneme 0.8.14 → 0.8.16

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 -21
  2. package/README.md +660 -654
  3. package/assets/icon.svg +3 -3
  4. package/bin/cli.mjs +691 -689
  5. package/bin/dsh-mneme-mcp.mjs +510 -510
  6. package/bin/dsh-mneme-serve.mjs +151 -151
  7. package/cordis.patch.yml +21 -21
  8. package/docs/CONFIGURATION.md +275 -0
  9. package/docs/CONTINUITY.md +124 -0
  10. package/docs/DAEMON.md +46 -0
  11. package/docs/DISTILL.md +162 -0
  12. package/docs/ENTITIES.md +245 -0
  13. package/docs/LOCAL_MODEL.md +222 -0
  14. package/docs/SEMANTIC.md +256 -0
  15. package/docs/SLEEP.md +164 -0
  16. package/docs/STORAGE.md +98 -0
  17. package/lib/api-standalone.js +656 -673
  18. package/lib/api.js +1605 -1433
  19. package/lib/bootstrap.js +211 -211
  20. package/lib/client.js +6354 -5296
  21. package/lib/commands.js +95 -95
  22. package/lib/config.js +947 -843
  23. package/lib/content-hash.js +54 -54
  24. package/lib/continuity.js +220 -220
  25. package/lib/document-index.js +94 -94
  26. package/lib/document.js +303 -303
  27. package/lib/dream/clustering.js +118 -118
  28. package/lib/dream/decisions.js +694 -674
  29. package/lib/dream/narratives.js +44 -44
  30. package/lib/dream/sleep.js +819 -819
  31. package/lib/dream.js +1503 -1501
  32. package/lib/embedding.js +177 -177
  33. package/lib/entities/extractor.js +252 -252
  34. package/lib/graph/anchoring.js +179 -179
  35. package/lib/guide.js +74 -74
  36. package/lib/heat.js +144 -144
  37. package/lib/hot-memory.js +108 -46
  38. package/lib/index.js +552 -503
  39. package/lib/inject.js +428 -450
  40. package/lib/lang.js +248 -229
  41. package/lib/local-embedder.js +358 -358
  42. package/lib/maintenance.js +162 -142
  43. package/lib/mirror.js +330 -330
  44. package/lib/organize.js +450 -450
  45. package/lib/peak-hours.js +117 -117
  46. package/lib/placement.js +13 -0
  47. package/lib/pre-inject-gate.js +340 -0
  48. package/lib/quality-filter.js +126 -126
  49. package/lib/recall-stats.js +190 -175
  50. package/lib/reclaim-http.js +150 -0
  51. package/lib/reranker.js +269 -269
  52. package/lib/runtime/adopt.js +176 -176
  53. package/lib/runtime/closure.js +172 -172
  54. package/lib/runtime/download.js +247 -247
  55. package/lib/runtime/layout.js +265 -265
  56. package/lib/runtime/loader.js +252 -252
  57. package/lib/runtime/model-fetch.js +361 -361
  58. package/lib/runtime/provision.js +237 -237
  59. package/lib/runtime/tarball.js +130 -130
  60. package/lib/runtime/verify.js +223 -204
  61. package/lib/safe-slice.js +18 -0
  62. package/lib/scope.js +146 -146
  63. package/lib/search/adaptive.js +60 -60
  64. package/lib/search/bm25.js +96 -96
  65. package/lib/semantic.js +229 -229
  66. package/lib/sensitive-scan.js +209 -209
  67. package/lib/serve.js +215 -215
  68. package/lib/service.js +2688 -2605
  69. package/lib/settings.js +528 -511
  70. package/lib/store.js +3196 -3007
  71. package/lib/summarize.js +1039 -856
  72. package/lib/tools.js +743 -737
  73. package/lib/vector-index.js +110 -110
  74. package/lib/version-check.js +122 -122
  75. package/lib/write-admission.js +402 -402
  76. package/locale/en.json +6 -6
  77. package/locale/zh.json +6 -6
  78. package/package.json +118 -118
  79. package/runtime-manifest.json +526 -526
  80. package/scripts/benchmark-embed.js +201 -201
  81. package/scripts/benchmark-recall.js +249 -249
  82. package/scripts/benchmark-rerank.js +166 -166
  83. package/scripts/build-runtime-manifest.mjs +224 -224
  84. package/scripts/changelog-prep.mjs +36 -36
  85. package/scripts/check-sync.js +42 -42
  86. package/scripts/e2e-dsh.js +220 -220
  87. package/scripts/mneme-runtime.mjs +277 -277
  88. package/scripts/stress-dsh.js +255 -255
  89. package/scripts/sync-lib.js +47 -47
  90. package/scripts/sync-test-badge.mjs +69 -69
  91. package/scripts/test-count-sync.mjs +44 -44
  92. package/src/api-standalone.js +656 -673
  93. package/src/api.js +1605 -1433
  94. package/src/bootstrap.js +211 -211
  95. package/src/commands.js +95 -95
  96. package/src/config.js +947 -843
  97. package/src/content-hash.js +54 -54
  98. package/src/continuity.js +220 -220
  99. package/src/document-index.js +94 -94
  100. package/src/document.js +303 -303
  101. package/src/dream/clustering.js +118 -118
  102. package/src/dream/decisions.js +694 -674
  103. package/src/dream/narratives.js +44 -44
  104. package/src/dream/sleep.js +819 -819
  105. package/src/dream.js +1503 -1501
  106. package/src/embedding.js +177 -177
  107. package/src/entities/extractor.js +252 -252
  108. package/src/graph/anchoring.js +179 -179
  109. package/src/guide.js +74 -74
  110. package/src/heat.js +144 -144
  111. package/src/hot-memory.js +108 -46
  112. package/src/index.js +552 -503
  113. package/src/inject.js +428 -450
  114. package/src/lang.js +248 -229
  115. package/src/local-embedder.js +358 -358
  116. package/src/maintenance.js +162 -142
  117. package/src/mirror.js +330 -330
  118. package/src/organize.js +450 -450
  119. package/src/peak-hours.js +117 -117
  120. package/src/placement.js +13 -0
  121. package/src/pre-inject-gate.js +340 -0
  122. package/src/quality-filter.js +126 -126
  123. package/src/recall-stats.js +190 -175
  124. package/src/reclaim-http.js +150 -0
  125. package/src/reranker.js +269 -269
  126. package/src/runtime/adopt.js +176 -176
  127. package/src/runtime/closure.js +172 -172
  128. package/src/runtime/download.js +247 -247
  129. package/src/runtime/layout.js +265 -265
  130. package/src/runtime/loader.js +252 -252
  131. package/src/runtime/model-fetch.js +361 -361
  132. package/src/runtime/provision.js +237 -237
  133. package/src/runtime/tarball.js +130 -130
  134. package/src/runtime/verify.js +223 -204
  135. package/src/safe-slice.js +18 -0
  136. package/src/scope.js +146 -146
  137. package/src/search/adaptive.js +60 -60
  138. package/src/search/bm25.js +96 -96
  139. package/src/semantic.js +229 -229
  140. package/src/sensitive-scan.js +209 -209
  141. package/src/serve.js +215 -215
  142. package/src/service.js +2688 -2605
  143. package/src/settings.js +528 -511
  144. package/src/store.js +3196 -3007
  145. package/src/summarize.js +1039 -856
  146. package/src/tools.js +743 -737
  147. package/src/vector-index.js +110 -110
  148. package/src/version-check.js +122 -122
  149. package/src/write-admission.js +402 -402
  150. package/test/adaptive-inject-budget.test.js +85 -85
  151. package/test/api-routes-gaps.test.js +899 -218
  152. package/test/api.test.js +1725 -1690
  153. package/test/audit.test.js +448 -448
  154. package/test/baseline-persist.test.js +248 -248
  155. package/test/benchmark.test.js +94 -94
  156. package/test/bootstrap.test.js +220 -220
  157. package/test/changelog-prep.test.js +60 -60
  158. package/test/client.test.js +1582 -1169
  159. package/test/clustering.test.js +100 -100
  160. package/test/commands.test.js +106 -106
  161. package/test/config.test.js +66 -66
  162. package/test/conflict-freeze.test.js +300 -300
  163. package/test/content-hash.test.js +204 -204
  164. package/test/continuity.test.js +333 -333
  165. package/test/contrast.test.js +314 -0
  166. package/test/document-dir.test.js +233 -233
  167. package/test/document.test.js +671 -671
  168. package/test/dream-archive-guardrail.test.js +116 -116
  169. package/test/dream-candidate-hybrid.test.js +281 -281
  170. package/test/dream-coverage-degrade.test.js +86 -86
  171. package/test/dream-failure-backoff.test.js +173 -173
  172. package/test/dream-merge-guardrail.test.js +219 -0
  173. package/test/dream-peak-hours.test.js +194 -194
  174. package/test/dream.test.js +1385 -1383
  175. package/test/embedding-ready.test.js +196 -196
  176. package/test/entities.test.js +627 -547
  177. package/test/entity-recall.test.js +179 -179
  178. package/test/epistemic.test.js +325 -325
  179. package/test/fnew-0112.test.js +311 -311
  180. package/test/fnew-03.test.js +422 -422
  181. package/test/graph-anchoring.test.js +184 -184
  182. package/test/graph-api.test.js +175 -175
  183. package/test/graph-inject-budget.test.js +103 -103
  184. package/test/graph-passive-confirm.test.js +91 -91
  185. package/test/graph-weight-evolution.test.js +104 -104
  186. package/test/heat.test.js +175 -175
  187. package/test/helpers/dream-mock.js +82 -82
  188. package/test/helpers/peer-worker.mjs +25 -25
  189. package/test/helpers/runtime-source.js +60 -60
  190. package/test/helpers/tar-builder.js +80 -80
  191. package/test/helpers/write-admission-samples.js +308 -308
  192. package/test/hot-memory.test.js +211 -145
  193. package/test/inject-content-cap.test.js +91 -91
  194. package/test/inject-first-round-bm25.test.js +89 -89
  195. package/test/inject-heat.test.js +77 -77
  196. package/test/inject-parent-gate.test.js +106 -106
  197. package/test/inject-pools.test.js +194 -194
  198. package/test/inject-recall.test.js +144 -105
  199. package/test/inject-rotation.test.js +161 -161
  200. package/test/inject.test.js +443 -420
  201. package/test/lang.test.js +87 -87
  202. package/test/lastrunat-persist.test.js +101 -101
  203. package/test/lib-smoke.test.js +96 -96
  204. package/test/llm-audit.test.js +696 -598
  205. package/test/llm-message-source.test.js +207 -207
  206. package/test/local-embedder.test.js +292 -292
  207. package/test/maintenance.test.js +501 -391
  208. package/test/mcp-stdio.test.js +385 -385
  209. package/test/mirror-dirty.test.js +424 -424
  210. package/test/mirror-edit-digest.test.js +187 -187
  211. package/test/mirror-generation.test.js +499 -499
  212. package/test/mirror-types.test.js +208 -208
  213. package/test/mirror.test.js +249 -249
  214. package/test/narrative-bars.test.js +162 -162
  215. package/test/organize.test.js +418 -418
  216. package/test/peer-blockers.test.js +193 -193
  217. package/test/placement.test.js +36 -0
  218. package/test/policy-epoch.test.js +259 -259
  219. package/test/pre-inject-gate.test.js +444 -0
  220. package/test/quality-filter.test.js +186 -186
  221. package/test/reasoning-effort.test.js +752 -752
  222. package/test/recall-anti-update.test.js +82 -82
  223. package/test/recall-evals.test.js +253 -253
  224. package/test/recall-layer.test.js +326 -326
  225. package/test/recall-stats.test.js +274 -189
  226. package/test/receipt-chain.test.js +451 -451
  227. package/test/reclaim-http.test.js +209 -0
  228. package/test/reflection.test.js +226 -226
  229. package/test/reindex-backfill.test.js +62 -62
  230. package/test/reranker-single-logit.test.js +73 -73
  231. package/test/reranker.test.js +197 -197
  232. package/test/resident-status.test.js +83 -83
  233. package/test/runtime-adopt.test.js +206 -206
  234. package/test/runtime-cli.test.js +89 -89
  235. package/test/runtime-closure.test.js +193 -193
  236. package/test/runtime-download.test.js +302 -302
  237. package/test/runtime-layout.test.js +163 -163
  238. package/test/runtime-loader.test.js +255 -255
  239. package/test/runtime-manifest.test.js +159 -159
  240. package/test/runtime-model-fetch.test.js +348 -348
  241. package/test/runtime-provision.test.js +101 -101
  242. package/test/runtime-tarball.test.js +81 -81
  243. package/test/runtime-verify.test.js +319 -279
  244. package/test/scope-candidates.test.js +239 -239
  245. package/test/scope-explicit.test.js +452 -452
  246. package/test/scope-inject-soft.test.js +106 -106
  247. package/test/scope-retrieval.test.js +197 -197
  248. package/test/scope-storage.test.js +255 -255
  249. package/test/scope-strict.test.js +367 -367
  250. package/test/search-fusion.test.js +90 -90
  251. package/test/semantic.test.js +124 -124
  252. package/test/sensitive-scan.test.js +207 -207
  253. package/test/serve-bin.test.js +121 -121
  254. package/test/serve.test.js +196 -196
  255. package/test/service-search.test.js +174 -174
  256. package/test/service.test.js +410 -405
  257. package/test/settings.test.js +269 -261
  258. package/test/skip-invalid-restore.test.js +262 -262
  259. package/test/sleep-action-set.test.js +422 -422
  260. package/test/sleep-heat.test.js +126 -126
  261. package/test/sleep-scheduler-rearm.test.js +89 -89
  262. package/test/sleep.test.js +527 -527
  263. package/test/standalone-api.test.js +751 -751
  264. package/test/store-vector-cache.test.js +98 -98
  265. package/test/store.test.js +504 -396
  266. package/test/stress.test.js +209 -209
  267. package/test/summarize-occurred-at.test.js +127 -0
  268. package/test/summarize-reasoning-effort.test.js +189 -189
  269. package/test/summarize.test.js +1639 -1465
  270. package/test/supersede-state-tracking.test.js +394 -0
  271. package/test/test-count-sync.test.js +60 -60
  272. package/test/tools-dto-schema-matrix.test.js +325 -325
  273. package/test/tools-memory-get-schema.test.js +65 -65
  274. package/test/tools.test.js +428 -374
  275. package/test/updated-at-semantics.test.js +113 -113
  276. package/test/vector-index.test.js +221 -221
  277. package/test/version-check.test.js +152 -152
  278. package/test/webServer-optional.test.js +82 -82
  279. package/test/write-admission.test.js +668 -668
@@ -1,247 +1,247 @@
1
- // 运行时取件(issue #131 / PR-C)
2
- //
3
- // 三档来源,按可用性依次尝试。第一档「收编本机 node_modules」在 adopt.js;这里管后两档:
4
- // ② 本地 .tgz 目录 —— 给「某个包网络下不到」的情况(本机就有一份 82MB 的 vendor tarball);
5
- // ③ registry(可换镜像)—— 有网络时的常规路径。
6
- //
7
- // 三条硬规矩:
8
- // 1. **先校验 sha512 再落盘**。不匹配就不写、整次判失败。没有「先信任再校验」这种中间态。
9
- // 2. **tar 条目路径必须落在该包目录内**。registry 与本地 tarball 都是外来输入,`../` 这类
10
- // 路径遍历要在这里挡住 —— 这是信任边界上的校验,不是美化。
11
- // 3. **任何一步失败都清掉半份 payload**。结构检查只看必需包与入口,半份 payload 可能
12
- // 「看起来能用」,然后在 import 原生模块时才炸,错误现场离真正的原因很远。
13
- //
14
- // @module dsh-mneme/runtime/download
15
- import { existsSync, mkdirSync, readFileSync, rmSync, writeFileSync } from "node:fs";
16
- import { dirname, join, sep } from "node:path";
17
- import { RUNTIME_MANIFEST_VERSION } from "./adopt.js";
18
- import { REQUIRED_PACKAGES, defaultRuntimeDir, detectLibc, nodeModulesDir, payloadDir, payloadId, runtimeManifestPath } from "./layout.js";
19
- import { matchesPlatform } from "./closure.js";
20
- import { stripPackagePrefix, verifyIntegrity, walkTgz } from "./tarball.js";
21
-
22
- const DEFAULT_ATTEMPTS = 3;
23
- // 单次请求的总超时。连接僵住(对端不回也不断)时,没有它就会永远等着 —— 用户看到的是一个
24
- // 卡死的按钮,而没有任何错误可看。给得宽松:几百 MB 的包在慢网下也要能下完。
25
- const DEFAULT_REQUEST_TIMEOUT_MS = 10 * 60 * 1000;
26
-
27
- /**
28
- * 把 registry 地址换成镜像地址:只换前缀,保留 `/name/-/file.tgz` 这段路径。
29
- * 换不了(不是 registry.npmjs.org 的地址)就原样返回 —— 清单里写死的地址优先。
30
- * @param {string} url - 清单里的地址。
31
- * @param {string} mirror - 镜像前缀,空 = 不换。
32
- * @returns {string} 实际要取的地址。
33
- */
34
- export function applyMirror(url, mirror) {
35
- const base = String(mirror ?? "").trim().replace(/\/+$/, "");
36
- if (base === "") return url;
37
- // 只换 registry.npmjs.org 的地址:别的来源(自建 registry、file:、清单里已是镜像的地址)一律
38
- // 原样返回。这里曾经用 indexOf(...) 直接加偏移量拼字符串 —— 地址里没有那个域名时 indexOf 返回
39
- // -1,slice 出的是垃圾名字,把一条本来可用的地址改坏。
40
- const prefix = "https://registry.npmjs.org/";
41
- const marker = url.indexOf("/-/");
42
- if (marker === -1 || !url.startsWith(prefix)) return url;
43
- const name = url.slice(prefix.length, marker);
44
- return name === "" ? url : `${base}/${name}${url.slice(marker)}`;
45
- }
46
-
47
- /**
48
- * 本地 tarball 目录里的候选文件名。用 npm 的约定(`<basename>-<version>.tgz`),
49
- * 这样
50
- pm pack` 出来的东西直接丢进目录就能用,不需要改名字。
51
- * @param {string} dir - 目录。
52
- * @param {{name: string, version: string}} pkg - 包。
53
- * @returns {string} 候选路径。
54
- */
55
- export function localTarballPath(dir, pkg) {
56
- const base = pkg.name.startsWith("@") ? pkg.name.slice(pkg.name.indexOf("/") + 1) : pkg.name;
57
- return join(dir, `${base}-${pkg.version}.tgz`);
58
- }
59
-
60
- /**
61
- * 取一个 .tgz 的字节。失败重试;已经收到部分字节时用 `Range` **续传**而不是从头再来
62
- * —— 82MB 的 onnxruntime-node 在移动网络上重来一次的代价太大。
63
- * @param {string} url - 地址。
64
- * @param {object} [opts] - 选项。
65
- * @returns {Promise<{buffer: Buffer, attempts: number, resumed: boolean}>}
66
- */
67
- export async function fetchTarball(url, { attempts = DEFAULT_ATTEMPTS, fetchImpl = fetch, onProgress, timeoutMs = DEFAULT_REQUEST_TIMEOUT_MS } = {}) {
68
- let lastError = null;
69
- let chunks = [];
70
- let received = 0;
71
-
72
- for (let attempt = 1; attempt <= attempts; attempt++) {
73
- try {
74
- const options = received > 0 ? { headers: { Range: `bytes=${received}-` } } : {};
75
- // AbortSignal.timeout 是**总时长**上限,不是空闲超时:宁可给得宽松,也不要掐断正常的慢速下载。
76
- if (Number.isFinite(timeoutMs) && timeoutMs > 0) options.signal = AbortSignal.timeout(timeoutMs);
77
- const res = await fetchImpl(url, Object.keys(options).length > 0 ? options : undefined);
78
- if (res.status === 206) {
79
- // 续传被接受:已有字节保留,接着追加。
80
- } else if (res.ok) {
81
- // 服务端不支持续传(或首次请求):从头来。
82
- chunks = [];
83
- received = 0;
84
- } else {
85
- throw new Error(`HTTP ${res.status}`);
86
- }
87
-
88
- if (res.body?.getReader === undefined) {
89
- const whole = Buffer.from(await res.arrayBuffer());
90
- // 206 的响应体只是**剩余**那段,必须追加;200 才是完整的一份 —— 覆盖会丢掉前半段,
91
- // 于是 sha512 校验必挂(而这次失败看起来像「来源不可信」,与真实原因不符)。
92
- if (res.status === 206) {
93
- chunks.push(whole);
94
- received += whole.length;
95
- } else {
96
- chunks = [whole];
97
- received = whole.length;
98
- }
99
- } else {
100
- const reader = res.body.getReader();
101
- for (;;) {
102
- const { done, value } = await reader.read();
103
- if (done) break;
104
- const chunk = Buffer.from(value);
105
- chunks.push(chunk);
106
- received += chunk.length;
107
- // 事件带 type:同一个回调同时承载「字节级」与「包级」两种进度,不标类型调用方就只能猜。
108
- onProgress?.({ type: "bytes", received, attempt });
109
- }
110
- }
111
- return { buffer: Buffer.concat(chunks, received), attempts: attempt, resumed: attempt > 1 };
112
- } catch (error) {
113
- lastError = error;
114
- }
115
- }
116
- throw new Error(`取件失败(试了 ${attempts} 次):${lastError?.message ?? lastError}`);
117
- }
118
-
119
- /** 解一个 .tgz 到该包的目录里;条目路径必须落在目录内。 */
120
- function extractInto(buffer, pkgDir, skipped) {
121
- const stats = { files: 0, bytes: 0 };
122
- const guard = pkgDir.endsWith(sep) ? pkgDir : `${pkgDir}${sep}`;
123
- const walked = walkTgz(buffer, (entry) => {
124
- const target = join(pkgDir, stripPackagePrefix(entry.path));
125
- // 越界即整次失败:把外来归档里的 ../ 写出去是不可接受的。
126
- if (!target.startsWith(guard)) throw new Error(`归档里有越出包目录的路径:${entry.path}`);
127
- mkdirSync(dirname(target), { recursive: true });
128
- writeFileSync(target, entry.data);
129
- stats.files++;
130
- stats.bytes += entry.data.length;
131
- });
132
- for (const item of walked.skipped) skipped.push(item);
133
- return stats;
134
- }
135
-
136
- /**
137
- * 按清单把运行时取回来并落盘(第二、三档来源)。
138
- * @param {object} opts - 选项。
139
- * @param {object} opts.manifest - runtime-manifest.json 的内容。
140
- * @param {string} [opts.runtimeDir] - 运行时根目录。
141
- * @param {string} [opts.localTarballDir] - 本地 .tgz 目录(有就用,优先于网络)。
142
- * @param {string} [opts.mirror] - registry 镜像前缀。
143
- * @param {boolean} [opts.overwrite] - 目标已存在时是否覆盖。
144
- * @returns {Promise<object>} 与 adoptRuntime 同形的结果;失败时 ok=false 且带 reason,不抛。
145
- */
146
- export async function downloadRuntime({
147
- manifest,
148
- platform = process.platform,
149
- arch = process.arch,
150
- libc = detectLibc(platform),
151
- runtimeDir = defaultRuntimeDir(),
152
- localTarballDir = "",
153
- mirror = "",
154
- overwrite = false,
155
- attempts = DEFAULT_ATTEMPTS,
156
- fetchImpl = fetch,
157
- onProgress
158
- }) {
159
- // v2 清单是**平台无关**的:里面是所有平台的包(各带 os / cpu / libc),这里按本机平台过滤。
160
- // 这样一份随包发布的静态清单就覆盖 win32 / darwin / linux × x64 / arm64 —— 不需要
161
- // 「谁在哪个系统上跑一次生成脚本」,也就不会让 macOS 用户卡在「清单里没有我的平台」。
162
- const all = Array.isArray(manifest?.packages) ? manifest.packages : null;
163
- if (all === null) return { ok: false, reason: "运行时清单不可用(缺少 packages 列表)" };
164
-
165
- const wanted = all.filter((pkg) => matchesPlatform(pkg, platform, arch, libc));
166
- const missingRequired = REQUIRED_PACKAGES.filter((name) => !wanted.some((pkg) => pkg.rel === name));
167
- if (missingRequired.length > 0) {
168
- // 目标平台在清单里就缺必需包时,现在失败好过装出一份「结构看着全、import 原生模块才炸」的
169
- // 运行时 —— 那时错误现场离原因已经很远。
170
- return { ok: false, reason: `清单里 ${platform}-${arch} 缺少必需包:${missingRequired.join("、")}` };
171
- }
172
-
173
- const id = payloadId({ version: manifest.transformersVersion, platform, arch });
174
-
175
- const dir = payloadDir(runtimeDir, id);
176
- if (existsSync(dir) && !overwrite) {
177
- return {
178
- ok: false,
179
- reason: `目标已存在:${dir}(需要覆盖请传 overwrite: true)`,
180
- payloadDir: dir,
181
- payloadId: id
182
- };
183
- }
184
- // 覆盖前先清空:否则上一次的残留文件会让结构检查看到「完整」,掩盖这一次的缺件。
185
- if (existsSync(dir)) rmSync(dir, { recursive: true, force: true });
186
-
187
- const totals = { packages: 0, files: 0, bytes: 0, fromLocal: 0, fromNetwork: 0 };
188
- const warnings = [];
189
- const packages = [];
190
- const skipped = [];
191
-
192
- try {
193
- for (const pkg of wanted) {
194
- const local = localTarballDir ? localTarballPath(localTarballDir, pkg) : "";
195
- const useLocal = local !== "" && existsSync(local);
196
- const url = applyMirror(pkg.tarball, mirror);
197
- const buffer = useLocal ? readFileSync(local) : (await fetchTarball(url, { attempts, fetchImpl, onProgress })).buffer;
198
- if (useLocal) totals.fromLocal++;
199
- else totals.fromNetwork++;
200
-
201
- if (!verifyIntegrity(buffer, pkg.integrity)) {
202
- // 校验失败必须停在这里:写下去就等于把「来源不可信」变成「已装好」。
203
- throw new Error(`${pkg.name}@${pkg.version} 的 sha512 与清单不符(来源:${useLocal ? local : url})`);
204
- }
205
-
206
- const stats = extractInto(buffer, join(nodeModulesDir(dir), pkg.rel), skipped);
207
- totals.files += stats.files;
208
- totals.bytes += stats.bytes;
209
- totals.packages++;
210
- packages.push({ name: pkg.name, version: pkg.version, rel: pkg.rel, optional: pkg.optional === true });
211
- onProgress?.({ type: "package", package: pkg.name, done: totals.packages, total: wanted.length });
212
- }
213
- } catch (error) {
214
- // 半份 payload 比没有更糟:清掉,让「没装成」这件事在磁盘上也成立。
215
- rmSync(dir, { recursive: true, force: true });
216
- return { ok: false, reason: String(error?.message ?? error) };
217
- }
218
-
219
- const written = {
220
- manifestVersion: RUNTIME_MANIFEST_VERSION,
221
- payloadId: id,
222
- entry: manifest.entry ?? "@huggingface/transformers",
223
- version: manifest.transformersVersion,
224
- platform,
225
- arch,
226
- procedure: "downloaded",
227
- // 每个 tarball 都与清单里的 sha512 比对过 —— 这是下载档独有的、可如实声明的来源完整性。
228
- integrity: { status: "verified", checked: packages.length, detail: "每个 tarball 的 sha512 与 runtime-manifest.json 一致" },
229
- createdAt: new Date().toISOString(),
230
- materialize: { mode: "download", linkError: null },
231
- sources: { mirror: mirror || null, localTarballDir: localTarballDir || null, counts: { local: totals.fromLocal, network: totals.fromNetwork } },
232
- packages,
233
- gaps: [],
234
- skipped,
235
- warnings
236
- };
237
- mkdirSync(dir, { recursive: true });
238
- writeFileSync(runtimeManifestPath(dir), `${JSON.stringify(written, null, 2)}\n`, "utf8");
239
- return {
240
- ok: true,
241
- payloadDir: dir,
242
- payloadId: id,
243
- version: manifest.transformersVersion,
244
- totals,
245
- manifest: written
246
- };
247
- }
1
+ // 运行时取件(issue #131 / PR-C)
2
+ //
3
+ // 三档来源,按可用性依次尝试。第一档「收编本机 node_modules」在 adopt.js;这里管后两档:
4
+ // ② 本地 .tgz 目录 —— 给「某个包网络下不到」的情况(本机就有一份 82MB 的 vendor tarball);
5
+ // ③ registry(可换镜像)—— 有网络时的常规路径。
6
+ //
7
+ // 三条硬规矩:
8
+ // 1. **先校验 sha512 再落盘**。不匹配就不写、整次判失败。没有「先信任再校验」这种中间态。
9
+ // 2. **tar 条目路径必须落在该包目录内**。registry 与本地 tarball 都是外来输入,`../` 这类
10
+ // 路径遍历要在这里挡住 —— 这是信任边界上的校验,不是美化。
11
+ // 3. **任何一步失败都清掉半份 payload**。结构检查只看必需包与入口,半份 payload 可能
12
+ // 「看起来能用」,然后在 import 原生模块时才炸,错误现场离真正的原因很远。
13
+ //
14
+ // @module dsh-mneme/runtime/download
15
+ import { existsSync, mkdirSync, readFileSync, rmSync, writeFileSync } from "node:fs";
16
+ import { dirname, join, sep } from "node:path";
17
+ import { RUNTIME_MANIFEST_VERSION } from "./adopt.js";
18
+ import { REQUIRED_PACKAGES, defaultRuntimeDir, detectLibc, nodeModulesDir, payloadDir, payloadId, runtimeManifestPath } from "./layout.js";
19
+ import { matchesPlatform } from "./closure.js";
20
+ import { stripPackagePrefix, verifyIntegrity, walkTgz } from "./tarball.js";
21
+
22
+ const DEFAULT_ATTEMPTS = 3;
23
+ // 单次请求的总超时。连接僵住(对端不回也不断)时,没有它就会永远等着 —— 用户看到的是一个
24
+ // 卡死的按钮,而没有任何错误可看。给得宽松:几百 MB 的包在慢网下也要能下完。
25
+ const DEFAULT_REQUEST_TIMEOUT_MS = 10 * 60 * 1000;
26
+
27
+ /**
28
+ * 把 registry 地址换成镜像地址:只换前缀,保留 `/name/-/file.tgz` 这段路径。
29
+ * 换不了(不是 registry.npmjs.org 的地址)就原样返回 —— 清单里写死的地址优先。
30
+ * @param {string} url - 清单里的地址。
31
+ * @param {string} mirror - 镜像前缀,空 = 不换。
32
+ * @returns {string} 实际要取的地址。
33
+ */
34
+ export function applyMirror(url, mirror) {
35
+ const base = String(mirror ?? "").trim().replace(/\/+$/, "");
36
+ if (base === "") return url;
37
+ // 只换 registry.npmjs.org 的地址:别的来源(自建 registry、file:、清单里已是镜像的地址)一律
38
+ // 原样返回。这里曾经用 indexOf(...) 直接加偏移量拼字符串 —— 地址里没有那个域名时 indexOf 返回
39
+ // -1,slice 出的是垃圾名字,把一条本来可用的地址改坏。
40
+ const prefix = "https://registry.npmjs.org/";
41
+ const marker = url.indexOf("/-/");
42
+ if (marker === -1 || !url.startsWith(prefix)) return url;
43
+ const name = url.slice(prefix.length, marker);
44
+ return name === "" ? url : `${base}/${name}${url.slice(marker)}`;
45
+ }
46
+
47
+ /**
48
+ * 本地 tarball 目录里的候选文件名。用 npm 的约定(`<basename>-<version>.tgz`),
49
+ * 这样
50
+ pm pack` 出来的东西直接丢进目录就能用,不需要改名字。
51
+ * @param {string} dir - 目录。
52
+ * @param {{name: string, version: string}} pkg - 包。
53
+ * @returns {string} 候选路径。
54
+ */
55
+ export function localTarballPath(dir, pkg) {
56
+ const base = pkg.name.startsWith("@") ? pkg.name.slice(pkg.name.indexOf("/") + 1) : pkg.name;
57
+ return join(dir, `${base}-${pkg.version}.tgz`);
58
+ }
59
+
60
+ /**
61
+ * 取一个 .tgz 的字节。失败重试;已经收到部分字节时用 `Range` **续传**而不是从头再来
62
+ * —— 82MB 的 onnxruntime-node 在移动网络上重来一次的代价太大。
63
+ * @param {string} url - 地址。
64
+ * @param {object} [opts] - 选项。
65
+ * @returns {Promise<{buffer: Buffer, attempts: number, resumed: boolean}>}
66
+ */
67
+ export async function fetchTarball(url, { attempts = DEFAULT_ATTEMPTS, fetchImpl = fetch, onProgress, timeoutMs = DEFAULT_REQUEST_TIMEOUT_MS } = {}) {
68
+ let lastError = null;
69
+ let chunks = [];
70
+ let received = 0;
71
+
72
+ for (let attempt = 1; attempt <= attempts; attempt++) {
73
+ try {
74
+ const options = received > 0 ? { headers: { Range: `bytes=${received}-` } } : {};
75
+ // AbortSignal.timeout 是**总时长**上限,不是空闲超时:宁可给得宽松,也不要掐断正常的慢速下载。
76
+ if (Number.isFinite(timeoutMs) && timeoutMs > 0) options.signal = AbortSignal.timeout(timeoutMs);
77
+ const res = await fetchImpl(url, Object.keys(options).length > 0 ? options : undefined);
78
+ if (res.status === 206) {
79
+ // 续传被接受:已有字节保留,接着追加。
80
+ } else if (res.ok) {
81
+ // 服务端不支持续传(或首次请求):从头来。
82
+ chunks = [];
83
+ received = 0;
84
+ } else {
85
+ throw new Error(`HTTP ${res.status}`);
86
+ }
87
+
88
+ if (res.body?.getReader === undefined) {
89
+ const whole = Buffer.from(await res.arrayBuffer());
90
+ // 206 的响应体只是**剩余**那段,必须追加;200 才是完整的一份 —— 覆盖会丢掉前半段,
91
+ // 于是 sha512 校验必挂(而这次失败看起来像「来源不可信」,与真实原因不符)。
92
+ if (res.status === 206) {
93
+ chunks.push(whole);
94
+ received += whole.length;
95
+ } else {
96
+ chunks = [whole];
97
+ received = whole.length;
98
+ }
99
+ } else {
100
+ const reader = res.body.getReader();
101
+ for (;;) {
102
+ const { done, value } = await reader.read();
103
+ if (done) break;
104
+ const chunk = Buffer.from(value);
105
+ chunks.push(chunk);
106
+ received += chunk.length;
107
+ // 事件带 type:同一个回调同时承载「字节级」与「包级」两种进度,不标类型调用方就只能猜。
108
+ onProgress?.({ type: "bytes", received, attempt });
109
+ }
110
+ }
111
+ return { buffer: Buffer.concat(chunks, received), attempts: attempt, resumed: attempt > 1 };
112
+ } catch (error) {
113
+ lastError = error;
114
+ }
115
+ }
116
+ throw new Error(`取件失败(试了 ${attempts} 次):${lastError?.message ?? lastError}`);
117
+ }
118
+
119
+ /** 解一个 .tgz 到该包的目录里;条目路径必须落在目录内。 */
120
+ function extractInto(buffer, pkgDir, skipped) {
121
+ const stats = { files: 0, bytes: 0 };
122
+ const guard = pkgDir.endsWith(sep) ? pkgDir : `${pkgDir}${sep}`;
123
+ const walked = walkTgz(buffer, (entry) => {
124
+ const target = join(pkgDir, stripPackagePrefix(entry.path));
125
+ // 越界即整次失败:把外来归档里的 ../ 写出去是不可接受的。
126
+ if (!target.startsWith(guard)) throw new Error(`归档里有越出包目录的路径:${entry.path}`);
127
+ mkdirSync(dirname(target), { recursive: true });
128
+ writeFileSync(target, entry.data);
129
+ stats.files++;
130
+ stats.bytes += entry.data.length;
131
+ });
132
+ for (const item of walked.skipped) skipped.push(item);
133
+ return stats;
134
+ }
135
+
136
+ /**
137
+ * 按清单把运行时取回来并落盘(第二、三档来源)。
138
+ * @param {object} opts - 选项。
139
+ * @param {object} opts.manifest - runtime-manifest.json 的内容。
140
+ * @param {string} [opts.runtimeDir] - 运行时根目录。
141
+ * @param {string} [opts.localTarballDir] - 本地 .tgz 目录(有就用,优先于网络)。
142
+ * @param {string} [opts.mirror] - registry 镜像前缀。
143
+ * @param {boolean} [opts.overwrite] - 目标已存在时是否覆盖。
144
+ * @returns {Promise<object>} 与 adoptRuntime 同形的结果;失败时 ok=false 且带 reason,不抛。
145
+ */
146
+ export async function downloadRuntime({
147
+ manifest,
148
+ platform = process.platform,
149
+ arch = process.arch,
150
+ libc = detectLibc(platform),
151
+ runtimeDir = defaultRuntimeDir(),
152
+ localTarballDir = "",
153
+ mirror = "",
154
+ overwrite = false,
155
+ attempts = DEFAULT_ATTEMPTS,
156
+ fetchImpl = fetch,
157
+ onProgress
158
+ }) {
159
+ // v2 清单是**平台无关**的:里面是所有平台的包(各带 os / cpu / libc),这里按本机平台过滤。
160
+ // 这样一份随包发布的静态清单就覆盖 win32 / darwin / linux × x64 / arm64 —— 不需要
161
+ // 「谁在哪个系统上跑一次生成脚本」,也就不会让 macOS 用户卡在「清单里没有我的平台」。
162
+ const all = Array.isArray(manifest?.packages) ? manifest.packages : null;
163
+ if (all === null) return { ok: false, reason: "运行时清单不可用(缺少 packages 列表)" };
164
+
165
+ const wanted = all.filter((pkg) => matchesPlatform(pkg, platform, arch, libc));
166
+ const missingRequired = REQUIRED_PACKAGES.filter((name) => !wanted.some((pkg) => pkg.rel === name));
167
+ if (missingRequired.length > 0) {
168
+ // 目标平台在清单里就缺必需包时,现在失败好过装出一份「结构看着全、import 原生模块才炸」的
169
+ // 运行时 —— 那时错误现场离原因已经很远。
170
+ return { ok: false, reason: `清单里 ${platform}-${arch} 缺少必需包:${missingRequired.join("、")}` };
171
+ }
172
+
173
+ const id = payloadId({ version: manifest.transformersVersion, platform, arch });
174
+
175
+ const dir = payloadDir(runtimeDir, id);
176
+ if (existsSync(dir) && !overwrite) {
177
+ return {
178
+ ok: false,
179
+ reason: `目标已存在:${dir}(需要覆盖请传 overwrite: true)`,
180
+ payloadDir: dir,
181
+ payloadId: id
182
+ };
183
+ }
184
+ // 覆盖前先清空:否则上一次的残留文件会让结构检查看到「完整」,掩盖这一次的缺件。
185
+ if (existsSync(dir)) rmSync(dir, { recursive: true, force: true });
186
+
187
+ const totals = { packages: 0, files: 0, bytes: 0, fromLocal: 0, fromNetwork: 0 };
188
+ const warnings = [];
189
+ const packages = [];
190
+ const skipped = [];
191
+
192
+ try {
193
+ for (const pkg of wanted) {
194
+ const local = localTarballDir ? localTarballPath(localTarballDir, pkg) : "";
195
+ const useLocal = local !== "" && existsSync(local);
196
+ const url = applyMirror(pkg.tarball, mirror);
197
+ const buffer = useLocal ? readFileSync(local) : (await fetchTarball(url, { attempts, fetchImpl, onProgress })).buffer;
198
+ if (useLocal) totals.fromLocal++;
199
+ else totals.fromNetwork++;
200
+
201
+ if (!verifyIntegrity(buffer, pkg.integrity)) {
202
+ // 校验失败必须停在这里:写下去就等于把「来源不可信」变成「已装好」。
203
+ throw new Error(`${pkg.name}@${pkg.version} 的 sha512 与清单不符(来源:${useLocal ? local : url})`);
204
+ }
205
+
206
+ const stats = extractInto(buffer, join(nodeModulesDir(dir), pkg.rel), skipped);
207
+ totals.files += stats.files;
208
+ totals.bytes += stats.bytes;
209
+ totals.packages++;
210
+ packages.push({ name: pkg.name, version: pkg.version, rel: pkg.rel, optional: pkg.optional === true });
211
+ onProgress?.({ type: "package", package: pkg.name, done: totals.packages, total: wanted.length });
212
+ }
213
+ } catch (error) {
214
+ // 半份 payload 比没有更糟:清掉,让「没装成」这件事在磁盘上也成立。
215
+ rmSync(dir, { recursive: true, force: true });
216
+ return { ok: false, reason: String(error?.message ?? error) };
217
+ }
218
+
219
+ const written = {
220
+ manifestVersion: RUNTIME_MANIFEST_VERSION,
221
+ payloadId: id,
222
+ entry: manifest.entry ?? "@huggingface/transformers",
223
+ version: manifest.transformersVersion,
224
+ platform,
225
+ arch,
226
+ procedure: "downloaded",
227
+ // 每个 tarball 都与清单里的 sha512 比对过 —— 这是下载档独有的、可如实声明的来源完整性。
228
+ integrity: { status: "verified", checked: packages.length, detail: "每个 tarball 的 sha512 与 runtime-manifest.json 一致" },
229
+ createdAt: new Date().toISOString(),
230
+ materialize: { mode: "download", linkError: null },
231
+ sources: { mirror: mirror || null, localTarballDir: localTarballDir || null, counts: { local: totals.fromLocal, network: totals.fromNetwork } },
232
+ packages,
233
+ gaps: [],
234
+ skipped,
235
+ warnings
236
+ };
237
+ mkdirSync(dir, { recursive: true });
238
+ writeFileSync(runtimeManifestPath(dir), `${JSON.stringify(written, null, 2)}\n`, "utf8");
239
+ return {
240
+ ok: true,
241
+ payloadDir: dir,
242
+ payloadId: id,
243
+ version: manifest.transformersVersion,
244
+ totals,
245
+ manifest: written
246
+ };
247
+ }