@modusensus/dsh-mneme 0.8.3 → 0.8.5

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 (205) hide show
  1. package/LICENSE +21 -21
  2. package/README.md +727 -636
  3. package/bin/cli.mjs +0 -0
  4. package/bin/dsh-mneme-mcp.mjs +510 -0
  5. package/cordis.patch.yml +21 -21
  6. package/lib/api-standalone.js +506 -316
  7. package/lib/api.js +1403 -1316
  8. package/lib/bootstrap.js +211 -0
  9. package/lib/client.js +4657 -4366
  10. package/lib/commands.js +95 -95
  11. package/lib/config.js +615 -510
  12. package/lib/document.js +221 -0
  13. package/lib/dream/clustering.js +118 -118
  14. package/lib/dream/decisions.js +658 -658
  15. package/lib/dream/narratives.js +44 -0
  16. package/lib/dream/sleep.js +702 -700
  17. package/lib/dream.js +147 -16
  18. package/lib/embedding.js +177 -177
  19. package/lib/entities/extractor.js +249 -249
  20. package/lib/guide.js +44 -0
  21. package/lib/heat.js +144 -136
  22. package/lib/hot-memory.js +46 -46
  23. package/lib/index.js +515 -515
  24. package/lib/inject.js +375 -295
  25. package/lib/lang.js +208 -179
  26. package/lib/local-embedder.js +320 -320
  27. package/lib/mirror.js +198 -198
  28. package/lib/organize.js +450 -0
  29. package/lib/quality-filter.js +126 -126
  30. package/lib/recall-stats.js +125 -0
  31. package/lib/reranker.js +269 -269
  32. package/lib/runtime/adopt.js +176 -176
  33. package/lib/runtime/closure.js +172 -172
  34. package/lib/runtime/download.js +247 -247
  35. package/lib/runtime/layout.js +265 -235
  36. package/lib/runtime/loader.js +252 -249
  37. package/lib/runtime/model-fetch.js +361 -361
  38. package/lib/runtime/provision.js +237 -237
  39. package/lib/runtime/tarball.js +130 -130
  40. package/lib/runtime/verify.js +204 -191
  41. package/lib/scope.js +118 -118
  42. package/lib/search/adaptive.js +60 -22
  43. package/lib/search/bm25.js +96 -96
  44. package/lib/service.js +2324 -2087
  45. package/lib/settings.js +459 -420
  46. package/lib/store.js +2549 -2367
  47. package/lib/summarize.js +820 -449
  48. package/lib/tools.js +702 -574
  49. package/lib/vector-index.js +110 -110
  50. package/lib/version-check.js +122 -122
  51. package/package.json +109 -108
  52. package/runtime-manifest.json +526 -596
  53. package/scripts/benchmark-embed.js +201 -201
  54. package/scripts/benchmark-recall.js +199 -199
  55. package/scripts/benchmark-rerank.js +166 -166
  56. package/scripts/build-runtime-manifest.mjs +224 -217
  57. package/scripts/check-sync.js +42 -42
  58. package/scripts/e2e-dsh.js +220 -220
  59. package/scripts/mneme-runtime.mjs +277 -277
  60. package/scripts/stress-dsh.js +255 -255
  61. package/scripts/sync-lib.js +47 -47
  62. package/scripts/sync-test-badge.mjs +67 -61
  63. package/src/api-standalone.js +506 -316
  64. package/src/api.js +1403 -1316
  65. package/src/bootstrap.js +211 -0
  66. package/src/commands.js +95 -95
  67. package/src/config.js +615 -510
  68. package/src/document.js +221 -0
  69. package/src/dream/clustering.js +118 -118
  70. package/src/dream/decisions.js +658 -658
  71. package/src/dream/narratives.js +44 -0
  72. package/src/dream/sleep.js +702 -700
  73. package/src/dream.js +147 -16
  74. package/src/embedding.js +177 -177
  75. package/src/entities/extractor.js +249 -249
  76. package/src/guide.js +44 -0
  77. package/src/heat.js +144 -136
  78. package/src/hot-memory.js +46 -46
  79. package/src/index.js +515 -515
  80. package/src/inject.js +375 -295
  81. package/src/lang.js +208 -179
  82. package/src/local-embedder.js +320 -320
  83. package/src/mirror.js +198 -198
  84. package/src/organize.js +450 -0
  85. package/src/quality-filter.js +126 -126
  86. package/src/recall-stats.js +125 -0
  87. package/src/reranker.js +269 -269
  88. package/src/runtime/adopt.js +176 -176
  89. package/src/runtime/closure.js +172 -172
  90. package/src/runtime/download.js +247 -247
  91. package/src/runtime/layout.js +265 -235
  92. package/src/runtime/loader.js +252 -249
  93. package/src/runtime/model-fetch.js +361 -361
  94. package/src/runtime/provision.js +237 -237
  95. package/src/runtime/tarball.js +130 -130
  96. package/src/runtime/verify.js +204 -191
  97. package/src/scope.js +118 -118
  98. package/src/search/adaptive.js +60 -22
  99. package/src/search/bm25.js +96 -96
  100. package/src/service.js +2324 -2087
  101. package/src/settings.js +459 -420
  102. package/src/store.js +2549 -2367
  103. package/src/summarize.js +820 -449
  104. package/src/tools.js +702 -574
  105. package/src/vector-index.js +110 -110
  106. package/src/version-check.js +122 -122
  107. package/test/adaptive-inject-budget.test.js +85 -0
  108. package/test/api-routes-gaps.test.js +218 -218
  109. package/test/api.test.js +1640 -1482
  110. package/test/audit.test.js +448 -448
  111. package/test/benchmark.test.js +94 -94
  112. package/test/bootstrap.test.js +220 -0
  113. package/test/client.test.js +860 -736
  114. package/test/clustering.test.js +100 -100
  115. package/test/commands.test.js +106 -106
  116. package/test/config.test.js +66 -66
  117. package/test/conflict-freeze.test.js +300 -300
  118. package/test/document.test.js +523 -0
  119. package/test/dream-archive-guardrail.test.js +116 -116
  120. package/test/dream-candidate-hybrid.test.js +281 -281
  121. package/test/dream-coverage-degrade.test.js +86 -86
  122. package/test/dream.test.js +1308 -1254
  123. package/test/embedding-ready.test.js +196 -196
  124. package/test/entities.test.js +547 -547
  125. package/test/entity-recall.test.js +179 -0
  126. package/test/epistemic.test.js +298 -298
  127. package/test/fnew-0112.test.js +311 -311
  128. package/test/fnew-03.test.js +422 -422
  129. package/test/graph-api.test.js +175 -175
  130. package/test/heat.test.js +175 -148
  131. package/test/helpers/dream-mock.js +82 -82
  132. package/test/helpers/peer-worker.mjs +25 -25
  133. package/test/helpers/runtime-source.js +60 -60
  134. package/test/helpers/tar-builder.js +80 -80
  135. package/test/hot-memory.test.js +145 -145
  136. package/test/inject-content-cap.test.js +91 -0
  137. package/test/inject-first-round-bm25.test.js +89 -89
  138. package/test/inject-heat.test.js +61 -0
  139. package/test/inject-pools.test.js +170 -0
  140. package/test/inject-recall.test.js +105 -0
  141. package/test/inject-rotation.test.js +161 -161
  142. package/test/inject.test.js +283 -249
  143. package/test/lang.test.js +87 -87
  144. package/test/lib-smoke.test.js +96 -96
  145. package/test/llm-audit.test.js +462 -309
  146. package/test/llm-message-source.test.js +177 -177
  147. package/test/local-embedder.test.js +249 -249
  148. package/test/mcp-stdio.test.js +385 -0
  149. package/test/mirror-dirty.test.js +424 -424
  150. package/test/mirror-edit-digest.test.js +187 -187
  151. package/test/mirror-generation.test.js +499 -499
  152. package/test/mirror.test.js +249 -249
  153. package/test/narrative-bars.test.js +162 -0
  154. package/test/organize.test.js +418 -0
  155. package/test/peer-blockers.test.js +193 -193
  156. package/test/policy-epoch.test.js +259 -259
  157. package/test/quality-filter.test.js +186 -186
  158. package/test/reasoning-effort.test.js +746 -746
  159. package/test/recall-evals.test.js +235 -235
  160. package/test/recall-layer.test.js +326 -326
  161. package/test/recall-stats.test.js +123 -0
  162. package/test/receipt-chain.test.js +451 -451
  163. package/test/reflection.test.js +226 -226
  164. package/test/reindex-backfill.test.js +62 -62
  165. package/test/reranker-single-logit.test.js +73 -73
  166. package/test/reranker.test.js +197 -197
  167. package/test/resident-status.test.js +83 -0
  168. package/test/runtime-adopt.test.js +206 -206
  169. package/test/runtime-cli.test.js +89 -89
  170. package/test/runtime-closure.test.js +193 -193
  171. package/test/runtime-download.test.js +302 -302
  172. package/test/runtime-layout.test.js +163 -163
  173. package/test/runtime-loader.test.js +255 -237
  174. package/test/runtime-manifest.test.js +159 -158
  175. package/test/runtime-model-fetch.test.js +348 -348
  176. package/test/runtime-provision.test.js +101 -101
  177. package/test/runtime-tarball.test.js +81 -81
  178. package/test/runtime-verify.test.js +279 -214
  179. package/test/scope-candidates.test.js +239 -239
  180. package/test/scope-explicit.test.js +452 -452
  181. package/test/scope-retrieval.test.js +197 -197
  182. package/test/scope-storage.test.js +255 -255
  183. package/test/scope-strict.test.js +231 -231
  184. package/test/search-fusion.test.js +90 -90
  185. package/test/semantic.test.js +124 -124
  186. package/test/service-search.test.js +174 -174
  187. package/test/service.test.js +405 -405
  188. package/test/settings.test.js +261 -261
  189. package/test/skip-invalid-restore.test.js +262 -262
  190. package/test/sleep-action-set.test.js +422 -422
  191. package/test/sleep-heat.test.js +126 -125
  192. package/test/sleep-scheduler-rearm.test.js +89 -89
  193. package/test/sleep.test.js +410 -373
  194. package/test/standalone-api.test.js +544 -359
  195. package/test/store-vector-cache.test.js +98 -98
  196. package/test/store.test.js +396 -383
  197. package/test/stress.test.js +209 -209
  198. package/test/summarize.test.js +1288 -575
  199. package/test/tools-dto-schema-matrix.test.js +325 -0
  200. package/test/tools-memory-get-schema.test.js +65 -65
  201. package/test/tools.test.js +374 -341
  202. package/test/updated-at-semantics.test.js +113 -113
  203. package/test/vector-index.test.js +221 -221
  204. package/test/version-check.test.js +152 -152
  205. package/test/webServer-optional.test.js +82 -82
@@ -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
+ }