@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.
- package/LICENSE +21 -21
- package/README.md +727 -636
- package/bin/cli.mjs +0 -0
- package/bin/dsh-mneme-mcp.mjs +510 -0
- package/cordis.patch.yml +21 -21
- package/lib/api-standalone.js +506 -316
- package/lib/api.js +1403 -1316
- package/lib/bootstrap.js +211 -0
- package/lib/client.js +4657 -4366
- package/lib/commands.js +95 -95
- package/lib/config.js +615 -510
- package/lib/document.js +221 -0
- package/lib/dream/clustering.js +118 -118
- package/lib/dream/decisions.js +658 -658
- package/lib/dream/narratives.js +44 -0
- package/lib/dream/sleep.js +702 -700
- package/lib/dream.js +147 -16
- package/lib/embedding.js +177 -177
- package/lib/entities/extractor.js +249 -249
- package/lib/guide.js +44 -0
- package/lib/heat.js +144 -136
- package/lib/hot-memory.js +46 -46
- package/lib/index.js +515 -515
- package/lib/inject.js +375 -295
- package/lib/lang.js +208 -179
- package/lib/local-embedder.js +320 -320
- package/lib/mirror.js +198 -198
- package/lib/organize.js +450 -0
- package/lib/quality-filter.js +126 -126
- package/lib/recall-stats.js +125 -0
- package/lib/reranker.js +269 -269
- package/lib/runtime/adopt.js +176 -176
- package/lib/runtime/closure.js +172 -172
- package/lib/runtime/download.js +247 -247
- package/lib/runtime/layout.js +265 -235
- package/lib/runtime/loader.js +252 -249
- package/lib/runtime/model-fetch.js +361 -361
- package/lib/runtime/provision.js +237 -237
- package/lib/runtime/tarball.js +130 -130
- package/lib/runtime/verify.js +204 -191
- package/lib/scope.js +118 -118
- package/lib/search/adaptive.js +60 -22
- package/lib/search/bm25.js +96 -96
- package/lib/service.js +2324 -2087
- package/lib/settings.js +459 -420
- package/lib/store.js +2549 -2367
- package/lib/summarize.js +820 -449
- package/lib/tools.js +702 -574
- package/lib/vector-index.js +110 -110
- package/lib/version-check.js +122 -122
- package/package.json +109 -108
- package/runtime-manifest.json +526 -596
- package/scripts/benchmark-embed.js +201 -201
- package/scripts/benchmark-recall.js +199 -199
- package/scripts/benchmark-rerank.js +166 -166
- package/scripts/build-runtime-manifest.mjs +224 -217
- package/scripts/check-sync.js +42 -42
- package/scripts/e2e-dsh.js +220 -220
- package/scripts/mneme-runtime.mjs +277 -277
- package/scripts/stress-dsh.js +255 -255
- package/scripts/sync-lib.js +47 -47
- package/scripts/sync-test-badge.mjs +67 -61
- package/src/api-standalone.js +506 -316
- package/src/api.js +1403 -1316
- package/src/bootstrap.js +211 -0
- package/src/commands.js +95 -95
- package/src/config.js +615 -510
- package/src/document.js +221 -0
- package/src/dream/clustering.js +118 -118
- package/src/dream/decisions.js +658 -658
- package/src/dream/narratives.js +44 -0
- package/src/dream/sleep.js +702 -700
- package/src/dream.js +147 -16
- package/src/embedding.js +177 -177
- package/src/entities/extractor.js +249 -249
- package/src/guide.js +44 -0
- package/src/heat.js +144 -136
- package/src/hot-memory.js +46 -46
- package/src/index.js +515 -515
- package/src/inject.js +375 -295
- package/src/lang.js +208 -179
- package/src/local-embedder.js +320 -320
- package/src/mirror.js +198 -198
- package/src/organize.js +450 -0
- package/src/quality-filter.js +126 -126
- package/src/recall-stats.js +125 -0
- package/src/reranker.js +269 -269
- package/src/runtime/adopt.js +176 -176
- package/src/runtime/closure.js +172 -172
- package/src/runtime/download.js +247 -247
- package/src/runtime/layout.js +265 -235
- package/src/runtime/loader.js +252 -249
- package/src/runtime/model-fetch.js +361 -361
- package/src/runtime/provision.js +237 -237
- package/src/runtime/tarball.js +130 -130
- package/src/runtime/verify.js +204 -191
- package/src/scope.js +118 -118
- package/src/search/adaptive.js +60 -22
- package/src/search/bm25.js +96 -96
- package/src/service.js +2324 -2087
- package/src/settings.js +459 -420
- package/src/store.js +2549 -2367
- package/src/summarize.js +820 -449
- package/src/tools.js +702 -574
- package/src/vector-index.js +110 -110
- package/src/version-check.js +122 -122
- package/test/adaptive-inject-budget.test.js +85 -0
- package/test/api-routes-gaps.test.js +218 -218
- package/test/api.test.js +1640 -1482
- package/test/audit.test.js +448 -448
- package/test/benchmark.test.js +94 -94
- package/test/bootstrap.test.js +220 -0
- package/test/client.test.js +860 -736
- package/test/clustering.test.js +100 -100
- package/test/commands.test.js +106 -106
- package/test/config.test.js +66 -66
- package/test/conflict-freeze.test.js +300 -300
- package/test/document.test.js +523 -0
- package/test/dream-archive-guardrail.test.js +116 -116
- package/test/dream-candidate-hybrid.test.js +281 -281
- package/test/dream-coverage-degrade.test.js +86 -86
- package/test/dream.test.js +1308 -1254
- package/test/embedding-ready.test.js +196 -196
- package/test/entities.test.js +547 -547
- package/test/entity-recall.test.js +179 -0
- package/test/epistemic.test.js +298 -298
- package/test/fnew-0112.test.js +311 -311
- package/test/fnew-03.test.js +422 -422
- package/test/graph-api.test.js +175 -175
- package/test/heat.test.js +175 -148
- package/test/helpers/dream-mock.js +82 -82
- package/test/helpers/peer-worker.mjs +25 -25
- package/test/helpers/runtime-source.js +60 -60
- package/test/helpers/tar-builder.js +80 -80
- package/test/hot-memory.test.js +145 -145
- package/test/inject-content-cap.test.js +91 -0
- package/test/inject-first-round-bm25.test.js +89 -89
- package/test/inject-heat.test.js +61 -0
- package/test/inject-pools.test.js +170 -0
- package/test/inject-recall.test.js +105 -0
- package/test/inject-rotation.test.js +161 -161
- package/test/inject.test.js +283 -249
- package/test/lang.test.js +87 -87
- package/test/lib-smoke.test.js +96 -96
- package/test/llm-audit.test.js +462 -309
- package/test/llm-message-source.test.js +177 -177
- package/test/local-embedder.test.js +249 -249
- package/test/mcp-stdio.test.js +385 -0
- package/test/mirror-dirty.test.js +424 -424
- package/test/mirror-edit-digest.test.js +187 -187
- package/test/mirror-generation.test.js +499 -499
- package/test/mirror.test.js +249 -249
- package/test/narrative-bars.test.js +162 -0
- package/test/organize.test.js +418 -0
- package/test/peer-blockers.test.js +193 -193
- package/test/policy-epoch.test.js +259 -259
- package/test/quality-filter.test.js +186 -186
- package/test/reasoning-effort.test.js +746 -746
- package/test/recall-evals.test.js +235 -235
- package/test/recall-layer.test.js +326 -326
- package/test/recall-stats.test.js +123 -0
- package/test/receipt-chain.test.js +451 -451
- package/test/reflection.test.js +226 -226
- package/test/reindex-backfill.test.js +62 -62
- package/test/reranker-single-logit.test.js +73 -73
- package/test/reranker.test.js +197 -197
- package/test/resident-status.test.js +83 -0
- package/test/runtime-adopt.test.js +206 -206
- package/test/runtime-cli.test.js +89 -89
- package/test/runtime-closure.test.js +193 -193
- package/test/runtime-download.test.js +302 -302
- package/test/runtime-layout.test.js +163 -163
- package/test/runtime-loader.test.js +255 -237
- package/test/runtime-manifest.test.js +159 -158
- package/test/runtime-model-fetch.test.js +348 -348
- package/test/runtime-provision.test.js +101 -101
- package/test/runtime-tarball.test.js +81 -81
- package/test/runtime-verify.test.js +279 -214
- package/test/scope-candidates.test.js +239 -239
- package/test/scope-explicit.test.js +452 -452
- package/test/scope-retrieval.test.js +197 -197
- package/test/scope-storage.test.js +255 -255
- package/test/scope-strict.test.js +231 -231
- package/test/search-fusion.test.js +90 -90
- package/test/semantic.test.js +124 -124
- package/test/service-search.test.js +174 -174
- package/test/service.test.js +405 -405
- package/test/settings.test.js +261 -261
- package/test/skip-invalid-restore.test.js +262 -262
- package/test/sleep-action-set.test.js +422 -422
- package/test/sleep-heat.test.js +126 -125
- package/test/sleep-scheduler-rearm.test.js +89 -89
- package/test/sleep.test.js +410 -373
- package/test/standalone-api.test.js +544 -359
- package/test/store-vector-cache.test.js +98 -98
- package/test/store.test.js +396 -383
- package/test/stress.test.js +209 -209
- package/test/summarize.test.js +1288 -575
- package/test/tools-dto-schema-matrix.test.js +325 -0
- package/test/tools-memory-get-schema.test.js +65 -65
- package/test/tools.test.js +374 -341
- package/test/updated-at-semantics.test.js +113 -113
- package/test/vector-index.test.js +221 -221
- package/test/version-check.test.js +152 -152
- package/test/webServer-optional.test.js +82 -82
|
@@ -1,80 +1,80 @@
|
|
|
1
|
-
// test/helpers/tar-builder.js
|
|
2
|
-
// 共享的 tar 造件器:tarball 读取层与取件层的测试都要造归档,放这里免得两份漂移。
|
|
3
|
-
// 只造真东西(POSIX ustar 布局 + 正确的校验和),不引 tar 库 —— 那正是被测代码要回避的依赖。
|
|
4
|
-
|
|
5
|
-
import { gzipSync } from "node:zlib";
|
|
6
|
-
|
|
7
|
-
/** 一个 ustar 头块;字段布局照 POSIX,校验和按规范先按空格算再写回。 */
|
|
8
|
-
export function tarHeader({ name = "", prefix = "", size = 0, type = "0" }) {
|
|
9
|
-
const header = Buffer.alloc(512);
|
|
10
|
-
header.write(name, 0, 100, "utf8");
|
|
11
|
-
header.write("0000644\0", 100, 8, "utf8");
|
|
12
|
-
header.write("0000000\0", 108, 8, "utf8");
|
|
13
|
-
header.write("0000000\0", 116, 8, "utf8");
|
|
14
|
-
header.write(`${size.toString(8).padStart(11, "0")}\0`, 124, 12, "utf8");
|
|
15
|
-
header.write("00000000000\0", 136, 12, "utf8");
|
|
16
|
-
header.write(" ", 148, 8, "utf8");
|
|
17
|
-
header.write(type, 156, 1, "utf8");
|
|
18
|
-
header.write("ustar\0", 257, 6, "utf8");
|
|
19
|
-
header.write("00", 263, 2, "utf8");
|
|
20
|
-
header.write(prefix, 345, 155, "utf8");
|
|
21
|
-
let sum = 0;
|
|
22
|
-
for (const byte of header) sum += byte;
|
|
23
|
-
header.write(`${sum.toString(8).padStart(6, "0")}\0 `, 148, 8, "utf8");
|
|
24
|
-
return header;
|
|
25
|
-
}
|
|
26
|
-
|
|
27
|
-
/** 普通文件条目(数据按 512 对齐补齐)。 */
|
|
28
|
-
export function tarFile(name, body, { prefix = "" } = {}) {
|
|
29
|
-
const data = Buffer.from(body, "utf8");
|
|
30
|
-
const padded = Buffer.alloc(Math.ceil(data.length / 512) * 512);
|
|
31
|
-
data.copy(padded);
|
|
32
|
-
return [tarHeader({ name, prefix, size: data.length, type: "0" }), padded];
|
|
33
|
-
}
|
|
34
|
-
|
|
35
|
-
/** 目录("5")/ 符号链接("2")等无数据条目。 */
|
|
36
|
-
export function tarMeta(name, type) {
|
|
37
|
-
return [tarHeader({ name, size: 0, type })];
|
|
38
|
-
}
|
|
39
|
-
|
|
40
|
-
/** pax 扩展头:载荷是 `长度 path=值\n`。 */
|
|
41
|
-
export function tarPax(path) {
|
|
42
|
-
const record = Buffer.from(`path=${path}\n`, "utf8");
|
|
43
|
-
const line = `${record.length + String(record.length).length} path=${path}\n`;
|
|
44
|
-
return [tarHeader({ name: "PaxHeaders/x", size: Buffer.byteLength(line), type: "x" }), padTo512(Buffer.from(line, "utf8"))];
|
|
45
|
-
}
|
|
46
|
-
|
|
47
|
-
/** 补齐到 512 的整数倍。 */
|
|
48
|
-
export function padTo512(buffer) {
|
|
49
|
-
if (buffer.length % 512 === 0) return buffer;
|
|
50
|
-
const padded = Buffer.alloc(Math.ceil(buffer.length / 512) * 512);
|
|
51
|
-
buffer.copy(padded);
|
|
52
|
-
return padded;
|
|
53
|
-
}
|
|
54
|
-
|
|
55
|
-
/** 两个全零块 = 归档结束。 */
|
|
56
|
-
export const END_BLOCKS = [Buffer.alloc(512), Buffer.alloc(512)];
|
|
57
|
-
|
|
58
|
-
/**
|
|
59
|
-
* 造一份 .tgz。`entries` 每项:
|
|
60
|
-
* { path, body } 普通文件(自动加 `package/` 前缀,除非 path 已含它)
|
|
61
|
-
* { path, type } 目录/符号链接等(type 是 ustar 的 typeflag 字符)
|
|
62
|
-
* { paxPath } pax 头,覆盖下一条目的路径
|
|
63
|
-
* @param {Array<object>} entries - 条目。
|
|
64
|
-
* @param {{prefix?: string|null, end?: boolean}} [opts] - 前缀(默认 package/)、是否追加结束块。
|
|
65
|
-
* @returns {Buffer} gzip 后的归档。
|
|
66
|
-
*/
|
|
67
|
-
export function makeTgz(entries, { prefix = "package/", end = true } = {}) {
|
|
68
|
-
const blocks = [];
|
|
69
|
-
for (const entry of entries) {
|
|
70
|
-
if (entry.paxPath !== undefined) {
|
|
71
|
-
blocks.push(...tarPax(prefix === null ? entry.paxPath : `${prefix}${entry.paxPath}`));
|
|
72
|
-
continue;
|
|
73
|
-
}
|
|
74
|
-
const path = prefix === null || entry.path.startsWith(prefix) ? entry.path : `${prefix}${entry.path}`;
|
|
75
|
-
if (entry.type !== undefined) blocks.push(...tarMeta(path, entry.type));
|
|
76
|
-
else blocks.push(...tarFile(path, entry.body ?? ""));
|
|
77
|
-
}
|
|
78
|
-
if (end) blocks.push(...END_BLOCKS);
|
|
79
|
-
return gzipSync(Buffer.concat(blocks));
|
|
80
|
-
}
|
|
1
|
+
// test/helpers/tar-builder.js
|
|
2
|
+
// 共享的 tar 造件器:tarball 读取层与取件层的测试都要造归档,放这里免得两份漂移。
|
|
3
|
+
// 只造真东西(POSIX ustar 布局 + 正确的校验和),不引 tar 库 —— 那正是被测代码要回避的依赖。
|
|
4
|
+
|
|
5
|
+
import { gzipSync } from "node:zlib";
|
|
6
|
+
|
|
7
|
+
/** 一个 ustar 头块;字段布局照 POSIX,校验和按规范先按空格算再写回。 */
|
|
8
|
+
export function tarHeader({ name = "", prefix = "", size = 0, type = "0" }) {
|
|
9
|
+
const header = Buffer.alloc(512);
|
|
10
|
+
header.write(name, 0, 100, "utf8");
|
|
11
|
+
header.write("0000644\0", 100, 8, "utf8");
|
|
12
|
+
header.write("0000000\0", 108, 8, "utf8");
|
|
13
|
+
header.write("0000000\0", 116, 8, "utf8");
|
|
14
|
+
header.write(`${size.toString(8).padStart(11, "0")}\0`, 124, 12, "utf8");
|
|
15
|
+
header.write("00000000000\0", 136, 12, "utf8");
|
|
16
|
+
header.write(" ", 148, 8, "utf8");
|
|
17
|
+
header.write(type, 156, 1, "utf8");
|
|
18
|
+
header.write("ustar\0", 257, 6, "utf8");
|
|
19
|
+
header.write("00", 263, 2, "utf8");
|
|
20
|
+
header.write(prefix, 345, 155, "utf8");
|
|
21
|
+
let sum = 0;
|
|
22
|
+
for (const byte of header) sum += byte;
|
|
23
|
+
header.write(`${sum.toString(8).padStart(6, "0")}\0 `, 148, 8, "utf8");
|
|
24
|
+
return header;
|
|
25
|
+
}
|
|
26
|
+
|
|
27
|
+
/** 普通文件条目(数据按 512 对齐补齐)。 */
|
|
28
|
+
export function tarFile(name, body, { prefix = "" } = {}) {
|
|
29
|
+
const data = Buffer.from(body, "utf8");
|
|
30
|
+
const padded = Buffer.alloc(Math.ceil(data.length / 512) * 512);
|
|
31
|
+
data.copy(padded);
|
|
32
|
+
return [tarHeader({ name, prefix, size: data.length, type: "0" }), padded];
|
|
33
|
+
}
|
|
34
|
+
|
|
35
|
+
/** 目录("5")/ 符号链接("2")等无数据条目。 */
|
|
36
|
+
export function tarMeta(name, type) {
|
|
37
|
+
return [tarHeader({ name, size: 0, type })];
|
|
38
|
+
}
|
|
39
|
+
|
|
40
|
+
/** pax 扩展头:载荷是 `长度 path=值\n`。 */
|
|
41
|
+
export function tarPax(path) {
|
|
42
|
+
const record = Buffer.from(`path=${path}\n`, "utf8");
|
|
43
|
+
const line = `${record.length + String(record.length).length} path=${path}\n`;
|
|
44
|
+
return [tarHeader({ name: "PaxHeaders/x", size: Buffer.byteLength(line), type: "x" }), padTo512(Buffer.from(line, "utf8"))];
|
|
45
|
+
}
|
|
46
|
+
|
|
47
|
+
/** 补齐到 512 的整数倍。 */
|
|
48
|
+
export function padTo512(buffer) {
|
|
49
|
+
if (buffer.length % 512 === 0) return buffer;
|
|
50
|
+
const padded = Buffer.alloc(Math.ceil(buffer.length / 512) * 512);
|
|
51
|
+
buffer.copy(padded);
|
|
52
|
+
return padded;
|
|
53
|
+
}
|
|
54
|
+
|
|
55
|
+
/** 两个全零块 = 归档结束。 */
|
|
56
|
+
export const END_BLOCKS = [Buffer.alloc(512), Buffer.alloc(512)];
|
|
57
|
+
|
|
58
|
+
/**
|
|
59
|
+
* 造一份 .tgz。`entries` 每项:
|
|
60
|
+
* { path, body } 普通文件(自动加 `package/` 前缀,除非 path 已含它)
|
|
61
|
+
* { path, type } 目录/符号链接等(type 是 ustar 的 typeflag 字符)
|
|
62
|
+
* { paxPath } pax 头,覆盖下一条目的路径
|
|
63
|
+
* @param {Array<object>} entries - 条目。
|
|
64
|
+
* @param {{prefix?: string|null, end?: boolean}} [opts] - 前缀(默认 package/)、是否追加结束块。
|
|
65
|
+
* @returns {Buffer} gzip 后的归档。
|
|
66
|
+
*/
|
|
67
|
+
export function makeTgz(entries, { prefix = "package/", end = true } = {}) {
|
|
68
|
+
const blocks = [];
|
|
69
|
+
for (const entry of entries) {
|
|
70
|
+
if (entry.paxPath !== undefined) {
|
|
71
|
+
blocks.push(...tarPax(prefix === null ? entry.paxPath : `${prefix}${entry.paxPath}`));
|
|
72
|
+
continue;
|
|
73
|
+
}
|
|
74
|
+
const path = prefix === null || entry.path.startsWith(prefix) ? entry.path : `${prefix}${entry.path}`;
|
|
75
|
+
if (entry.type !== undefined) blocks.push(...tarMeta(path, entry.type));
|
|
76
|
+
else blocks.push(...tarFile(path, entry.body ?? ""));
|
|
77
|
+
}
|
|
78
|
+
if (end) blocks.push(...END_BLOCKS);
|
|
79
|
+
return gzipSync(Buffer.concat(blocks));
|
|
80
|
+
}
|
package/test/hot-memory.test.js
CHANGED
|
@@ -1,145 +1,145 @@
|
|
|
1
|
-
import test from "node:test";
|
|
2
|
-
import assert from "node:assert/strict";
|
|
3
|
-
import { createHotMemory, estimateTokens } from "../src/hot-memory.js";
|
|
4
|
-
import { createStore } from "../src/store.js";
|
|
5
|
-
import { createService } from "../src/service.js";
|
|
6
|
-
import { createVectorIndex } from "../src/vector-index.js";
|
|
7
|
-
|
|
8
|
-
// --- hot memory buffer ---
|
|
9
|
-
|
|
10
|
-
test("hot memory keeps the latest rounds within maxRounds", () => {
|
|
11
|
-
const hot = createHotMemory({ maxRounds: 2, maxTokens: 10000 });
|
|
12
|
-
hot.add({ query: "第一轮", response: "答一" });
|
|
13
|
-
hot.add({ query: "第二轮", response: "答二" });
|
|
14
|
-
hot.add({ query: "第三轮", response: "答三" });
|
|
15
|
-
assert.equal(hot.rounds().length, 2);
|
|
16
|
-
assert.ok(hot.getContext().includes("第三轮"));
|
|
17
|
-
assert.ok(!hot.getContext().includes("第一轮"));
|
|
18
|
-
});
|
|
19
|
-
|
|
20
|
-
test("hot memory enforces the token budget", () => {
|
|
21
|
-
const hot = createHotMemory({ maxRounds: 10, maxTokens: 30 });
|
|
22
|
-
hot.add({ query: "很长的第一轮问题".repeat(10), response: "很长的回答".repeat(10) });
|
|
23
|
-
hot.add({ query: "第二轮", response: "答二" });
|
|
24
|
-
// The first round alone blows the budget; the newest round survives and
|
|
25
|
-
// the buffer never empties completely.
|
|
26
|
-
const rounds = hot.rounds();
|
|
27
|
-
assert.ok(rounds.length >= 1);
|
|
28
|
-
assert.equal(rounds[rounds.length - 1].query, "第二轮");
|
|
29
|
-
});
|
|
30
|
-
|
|
31
|
-
test("hot memory getContext uses the Q/A round format", () => {
|
|
32
|
-
const hot = createHotMemory({ maxRounds: 5, maxTokens: 10000 });
|
|
33
|
-
hot.add({ query: "Q1", response: "A1" });
|
|
34
|
-
hot.add({ query: "Q2", response: "A2" });
|
|
35
|
-
assert.equal(hot.getContext(), "Q: Q1\nA: A1\n\nQ: Q2\nA: A2");
|
|
36
|
-
});
|
|
37
|
-
|
|
38
|
-
test("hot memory ignores empty rounds and clears", () => {
|
|
39
|
-
const hot = createHotMemory({ maxRounds: 5, maxTokens: 10000 });
|
|
40
|
-
hot.add({ query: "", response: "x" });
|
|
41
|
-
hot.add(null);
|
|
42
|
-
assert.equal(hot.rounds().length, 0);
|
|
43
|
-
hot.add({ query: "q" });
|
|
44
|
-
hot.clear();
|
|
45
|
-
assert.equal(hot.getContext(), "");
|
|
46
|
-
});
|
|
47
|
-
|
|
48
|
-
test("estimateTokens counts CJK heavier than ASCII", () => {
|
|
49
|
-
assert.ok(estimateTokens("中文内容") > estimateTokens("abcd"));
|
|
50
|
-
});
|
|
51
|
-
|
|
52
|
-
// --- service-level: BM25 fusion + semantic dedup + selective injection ---
|
|
53
|
-
|
|
54
|
-
function toyVec(text) {
|
|
55
|
-
const v = new Array(64).fill(0);
|
|
56
|
-
for (const t of String(text).toLowerCase().split(/[^\p{L}\p{N}]+/u).filter(Boolean)) {
|
|
57
|
-
let h = 0;
|
|
58
|
-
for (const ch of t) h = (h * 31 + ch.codePointAt(0)) >>> 0;
|
|
59
|
-
v[h % 64] += 1;
|
|
60
|
-
}
|
|
61
|
-
const norm = Math.sqrt(v.reduce((s, x) => s + x * x, 0)) || 1;
|
|
62
|
-
return v.map((x) => x / norm);
|
|
63
|
-
}
|
|
64
|
-
|
|
65
|
-
function setup(overrides = {}) {
|
|
66
|
-
const store = createStore(":memory:");
|
|
67
|
-
const config = {
|
|
68
|
-
entitySearchEnabled: false,
|
|
69
|
-
bm25SearchEnabled: true,
|
|
70
|
-
adaptiveThresholdEnabled: false,
|
|
71
|
-
searchSemanticDedup: true,
|
|
72
|
-
selectiveInjectEnabled: true,
|
|
73
|
-
...overrides
|
|
74
|
-
};
|
|
75
|
-
const service = createService({ store, mirror: null, config, logger: null });
|
|
76
|
-
service.setVectorIndex(createVectorIndex({ store, logger: null }));
|
|
77
|
-
service.setEmbedder({ embedSingle: async (t) => toyVec(t) });
|
|
78
|
-
return { store, service };
|
|
79
|
-
}
|
|
80
|
-
|
|
81
|
-
test("searchMemories fuses BM25: scattered-term queries recall both rows", async () => {
|
|
82
|
-
const { store, service } = setup();
|
|
83
|
-
const a = store.save({ type: "project", title: "异步并发模式", content: "async runtime 选用 tokio", importance: 3 });
|
|
84
|
-
const b = store.save({ type: "decision", title: "语言迁移", content: "编译模块迁移到 Rust", importance: 3 });
|
|
85
|
-
store.save({ type: "project", title: "无关", content: "夜间 ETL 脚本", importance: 3 });
|
|
86
|
-
|
|
87
|
-
// "rust 异步" is not a substring of either row — LIKE misses both; BM25
|
|
88
|
-
// must surface both rows in the merged result.
|
|
89
|
-
const results = await service.searchMemories("rust 异步", { mode: "auto", topK: 5 });
|
|
90
|
-
const ids = results.map((r) => r.id);
|
|
91
|
-
assert.ok(ids.includes(a.id), "async row must be recalled via BM25");
|
|
92
|
-
assert.ok(ids.includes(b.id), "rust row must be recalled via BM25");
|
|
93
|
-
assert.equal(results.find((r) => r.id === a.id)?.source, "bm25");
|
|
94
|
-
});
|
|
95
|
-
|
|
96
|
-
test("bm25SearchEnabled=false restores the two-path behavior", async () => {
|
|
97
|
-
const { store, service } = setup({ bm25SearchEnabled: false });
|
|
98
|
-
const a = store.save({ type: "project", title: "异步并发模式", content: "async runtime 选用 tokio", importance: 3 });
|
|
99
|
-
const results = await service.searchMemories("rust 异步", { mode: "auto", topK: 5 });
|
|
100
|
-
assert.ok(!results.some((r) => r.id === a.id), "no BM25 → scattered-term miss is back");
|
|
101
|
-
});
|
|
102
|
-
|
|
103
|
-
test("search-time semantic dedup drops near-identical embeddings", async () => {
|
|
104
|
-
const { store, service } = setup();
|
|
105
|
-
const a = store.save({ type: "project", title: "偏好 A", content: "用户喜欢深色主题编辑器", importance: 3 });
|
|
106
|
-
const b = store.save({ type: "project", title: "偏好 A 备份", content: "用户喜欢深色主题编辑器(备份)", importance: 3 });
|
|
107
|
-
store.setEmbedding(a.id, toyVec("用户喜欢深色主题编辑器"));
|
|
108
|
-
store.setEmbedding(b.id, toyVec("用户喜欢深色主题编辑器"));
|
|
109
|
-
|
|
110
|
-
// auto (not keyword): keyword mode is the documented text-only path and is
|
|
111
|
-
// exempt from dedup by design; auto exercises the dedup the way production
|
|
112
|
-
// searches run.
|
|
113
|
-
const results = await service.searchMemories("深色主题", { mode: "auto", topK: 5 });
|
|
114
|
-
const ids = results.map((r) => r.id);
|
|
115
|
-
assert.ok(ids.includes(a.id) !== ids.includes(b.id), "one of the near-duplicate pair is dropped");
|
|
116
|
-
});
|
|
117
|
-
|
|
118
|
-
test("searchSemanticDedup=false keeps duplicate embeddings", async () => {
|
|
119
|
-
const { store, service } = setup({ searchSemanticDedup: false });
|
|
120
|
-
const a = store.save({ type: "project", title: "偏好 A", content: "用户喜欢深色主题编辑器", importance: 3 });
|
|
121
|
-
const b = store.save({ type: "project", title: "偏好 A 备份", content: "用户喜欢深色主题编辑器(备份)", importance: 3 });
|
|
122
|
-
store.setEmbedding(a.id, toyVec("用户喜欢深色主题编辑器"));
|
|
123
|
-
store.setEmbedding(b.id, toyVec("用户喜欢深色主题编辑器"));
|
|
124
|
-
const results = await service.searchMemories("深色主题", { mode: "keyword", topK: 5 });
|
|
125
|
-
assert.equal(results.length, 2);
|
|
126
|
-
});
|
|
127
|
-
|
|
128
|
-
test("selective injection re-orders candidates by query similarity", () => {
|
|
129
|
-
const { store, service } = setup();
|
|
130
|
-
const thesis = store.save({ type: "project", title: "湿地论文", content: "毕业论文研究城市湿地公园", importance: 5 });
|
|
131
|
-
const plugin = store.save({ type: "project", title: "插件项目", content: "dsh-mneme 记忆插件开发", importance: 5 });
|
|
132
|
-
store.setEmbedding(thesis.id, toyVec("毕业论文研究城市湿地公园"));
|
|
133
|
-
store.setEmbedding(plugin.id, toyVec("dsh-mneme 记忆插件开发"));
|
|
134
|
-
|
|
135
|
-
// Rule-based order would put both at equal importance; the query vector is
|
|
136
|
-
// about the thesis, so topic ranking must put the thesis memory first.
|
|
137
|
-
const picked = service.injectCandidates({
|
|
138
|
-
query: "论文写作",
|
|
139
|
-
queryVector: toyVec("毕业论文研究城市湿地公园"),
|
|
140
|
-
maxItems: 2,
|
|
141
|
-
threshold: 3
|
|
142
|
-
});
|
|
143
|
-
assert.equal(picked[0].id, thesis.id);
|
|
144
|
-
assert.ok(picked.some((m) => m.id === plugin.id));
|
|
145
|
-
});
|
|
1
|
+
import test from "node:test";
|
|
2
|
+
import assert from "node:assert/strict";
|
|
3
|
+
import { createHotMemory, estimateTokens } from "../src/hot-memory.js";
|
|
4
|
+
import { createStore } from "../src/store.js";
|
|
5
|
+
import { createService } from "../src/service.js";
|
|
6
|
+
import { createVectorIndex } from "../src/vector-index.js";
|
|
7
|
+
|
|
8
|
+
// --- hot memory buffer ---
|
|
9
|
+
|
|
10
|
+
test("hot memory keeps the latest rounds within maxRounds", () => {
|
|
11
|
+
const hot = createHotMemory({ maxRounds: 2, maxTokens: 10000 });
|
|
12
|
+
hot.add({ query: "第一轮", response: "答一" });
|
|
13
|
+
hot.add({ query: "第二轮", response: "答二" });
|
|
14
|
+
hot.add({ query: "第三轮", response: "答三" });
|
|
15
|
+
assert.equal(hot.rounds().length, 2);
|
|
16
|
+
assert.ok(hot.getContext().includes("第三轮"));
|
|
17
|
+
assert.ok(!hot.getContext().includes("第一轮"));
|
|
18
|
+
});
|
|
19
|
+
|
|
20
|
+
test("hot memory enforces the token budget", () => {
|
|
21
|
+
const hot = createHotMemory({ maxRounds: 10, maxTokens: 30 });
|
|
22
|
+
hot.add({ query: "很长的第一轮问题".repeat(10), response: "很长的回答".repeat(10) });
|
|
23
|
+
hot.add({ query: "第二轮", response: "答二" });
|
|
24
|
+
// The first round alone blows the budget; the newest round survives and
|
|
25
|
+
// the buffer never empties completely.
|
|
26
|
+
const rounds = hot.rounds();
|
|
27
|
+
assert.ok(rounds.length >= 1);
|
|
28
|
+
assert.equal(rounds[rounds.length - 1].query, "第二轮");
|
|
29
|
+
});
|
|
30
|
+
|
|
31
|
+
test("hot memory getContext uses the Q/A round format", () => {
|
|
32
|
+
const hot = createHotMemory({ maxRounds: 5, maxTokens: 10000 });
|
|
33
|
+
hot.add({ query: "Q1", response: "A1" });
|
|
34
|
+
hot.add({ query: "Q2", response: "A2" });
|
|
35
|
+
assert.equal(hot.getContext(), "Q: Q1\nA: A1\n\nQ: Q2\nA: A2");
|
|
36
|
+
});
|
|
37
|
+
|
|
38
|
+
test("hot memory ignores empty rounds and clears", () => {
|
|
39
|
+
const hot = createHotMemory({ maxRounds: 5, maxTokens: 10000 });
|
|
40
|
+
hot.add({ query: "", response: "x" });
|
|
41
|
+
hot.add(null);
|
|
42
|
+
assert.equal(hot.rounds().length, 0);
|
|
43
|
+
hot.add({ query: "q" });
|
|
44
|
+
hot.clear();
|
|
45
|
+
assert.equal(hot.getContext(), "");
|
|
46
|
+
});
|
|
47
|
+
|
|
48
|
+
test("estimateTokens counts CJK heavier than ASCII", () => {
|
|
49
|
+
assert.ok(estimateTokens("中文内容") > estimateTokens("abcd"));
|
|
50
|
+
});
|
|
51
|
+
|
|
52
|
+
// --- service-level: BM25 fusion + semantic dedup + selective injection ---
|
|
53
|
+
|
|
54
|
+
function toyVec(text) {
|
|
55
|
+
const v = new Array(64).fill(0);
|
|
56
|
+
for (const t of String(text).toLowerCase().split(/[^\p{L}\p{N}]+/u).filter(Boolean)) {
|
|
57
|
+
let h = 0;
|
|
58
|
+
for (const ch of t) h = (h * 31 + ch.codePointAt(0)) >>> 0;
|
|
59
|
+
v[h % 64] += 1;
|
|
60
|
+
}
|
|
61
|
+
const norm = Math.sqrt(v.reduce((s, x) => s + x * x, 0)) || 1;
|
|
62
|
+
return v.map((x) => x / norm);
|
|
63
|
+
}
|
|
64
|
+
|
|
65
|
+
function setup(overrides = {}) {
|
|
66
|
+
const store = createStore(":memory:");
|
|
67
|
+
const config = {
|
|
68
|
+
entitySearchEnabled: false,
|
|
69
|
+
bm25SearchEnabled: true,
|
|
70
|
+
adaptiveThresholdEnabled: false,
|
|
71
|
+
searchSemanticDedup: true,
|
|
72
|
+
selectiveInjectEnabled: true,
|
|
73
|
+
...overrides
|
|
74
|
+
};
|
|
75
|
+
const service = createService({ store, mirror: null, config, logger: null });
|
|
76
|
+
service.setVectorIndex(createVectorIndex({ store, logger: null }));
|
|
77
|
+
service.setEmbedder({ embedSingle: async (t) => toyVec(t) });
|
|
78
|
+
return { store, service };
|
|
79
|
+
}
|
|
80
|
+
|
|
81
|
+
test("searchMemories fuses BM25: scattered-term queries recall both rows", async () => {
|
|
82
|
+
const { store, service } = setup();
|
|
83
|
+
const a = store.save({ type: "project", title: "异步并发模式", content: "async runtime 选用 tokio", importance: 3 });
|
|
84
|
+
const b = store.save({ type: "decision", title: "语言迁移", content: "编译模块迁移到 Rust", importance: 3 });
|
|
85
|
+
store.save({ type: "project", title: "无关", content: "夜间 ETL 脚本", importance: 3 });
|
|
86
|
+
|
|
87
|
+
// "rust 异步" is not a substring of either row — LIKE misses both; BM25
|
|
88
|
+
// must surface both rows in the merged result.
|
|
89
|
+
const results = await service.searchMemories("rust 异步", { mode: "auto", topK: 5 });
|
|
90
|
+
const ids = results.map((r) => r.id);
|
|
91
|
+
assert.ok(ids.includes(a.id), "async row must be recalled via BM25");
|
|
92
|
+
assert.ok(ids.includes(b.id), "rust row must be recalled via BM25");
|
|
93
|
+
assert.equal(results.find((r) => r.id === a.id)?.source, "bm25");
|
|
94
|
+
});
|
|
95
|
+
|
|
96
|
+
test("bm25SearchEnabled=false restores the two-path behavior", async () => {
|
|
97
|
+
const { store, service } = setup({ bm25SearchEnabled: false });
|
|
98
|
+
const a = store.save({ type: "project", title: "异步并发模式", content: "async runtime 选用 tokio", importance: 3 });
|
|
99
|
+
const results = await service.searchMemories("rust 异步", { mode: "auto", topK: 5 });
|
|
100
|
+
assert.ok(!results.some((r) => r.id === a.id), "no BM25 → scattered-term miss is back");
|
|
101
|
+
});
|
|
102
|
+
|
|
103
|
+
test("search-time semantic dedup drops near-identical embeddings", async () => {
|
|
104
|
+
const { store, service } = setup();
|
|
105
|
+
const a = store.save({ type: "project", title: "偏好 A", content: "用户喜欢深色主题编辑器", importance: 3 });
|
|
106
|
+
const b = store.save({ type: "project", title: "偏好 A 备份", content: "用户喜欢深色主题编辑器(备份)", importance: 3 });
|
|
107
|
+
store.setEmbedding(a.id, toyVec("用户喜欢深色主题编辑器"));
|
|
108
|
+
store.setEmbedding(b.id, toyVec("用户喜欢深色主题编辑器"));
|
|
109
|
+
|
|
110
|
+
// auto (not keyword): keyword mode is the documented text-only path and is
|
|
111
|
+
// exempt from dedup by design; auto exercises the dedup the way production
|
|
112
|
+
// searches run.
|
|
113
|
+
const results = await service.searchMemories("深色主题", { mode: "auto", topK: 5 });
|
|
114
|
+
const ids = results.map((r) => r.id);
|
|
115
|
+
assert.ok(ids.includes(a.id) !== ids.includes(b.id), "one of the near-duplicate pair is dropped");
|
|
116
|
+
});
|
|
117
|
+
|
|
118
|
+
test("searchSemanticDedup=false keeps duplicate embeddings", async () => {
|
|
119
|
+
const { store, service } = setup({ searchSemanticDedup: false });
|
|
120
|
+
const a = store.save({ type: "project", title: "偏好 A", content: "用户喜欢深色主题编辑器", importance: 3 });
|
|
121
|
+
const b = store.save({ type: "project", title: "偏好 A 备份", content: "用户喜欢深色主题编辑器(备份)", importance: 3 });
|
|
122
|
+
store.setEmbedding(a.id, toyVec("用户喜欢深色主题编辑器"));
|
|
123
|
+
store.setEmbedding(b.id, toyVec("用户喜欢深色主题编辑器"));
|
|
124
|
+
const results = await service.searchMemories("深色主题", { mode: "keyword", topK: 5 });
|
|
125
|
+
assert.equal(results.length, 2);
|
|
126
|
+
});
|
|
127
|
+
|
|
128
|
+
test("selective injection re-orders candidates by query similarity", () => {
|
|
129
|
+
const { store, service } = setup();
|
|
130
|
+
const thesis = store.save({ type: "project", title: "湿地论文", content: "毕业论文研究城市湿地公园", importance: 5 });
|
|
131
|
+
const plugin = store.save({ type: "project", title: "插件项目", content: "dsh-mneme 记忆插件开发", importance: 5 });
|
|
132
|
+
store.setEmbedding(thesis.id, toyVec("毕业论文研究城市湿地公园"));
|
|
133
|
+
store.setEmbedding(plugin.id, toyVec("dsh-mneme 记忆插件开发"));
|
|
134
|
+
|
|
135
|
+
// Rule-based order would put both at equal importance; the query vector is
|
|
136
|
+
// about the thesis, so topic ranking must put the thesis memory first.
|
|
137
|
+
const picked = service.injectCandidates({
|
|
138
|
+
query: "论文写作",
|
|
139
|
+
queryVector: toyVec("毕业论文研究城市湿地公园"),
|
|
140
|
+
maxItems: 2,
|
|
141
|
+
threshold: 3
|
|
142
|
+
});
|
|
143
|
+
assert.equal(picked[0].id, thesis.id);
|
|
144
|
+
assert.ok(picked.some((m) => m.id === plugin.id));
|
|
145
|
+
});
|
|
@@ -0,0 +1,91 @@
|
|
|
1
|
+
// Issue #164① 注入截断修复测试。
|
|
2
|
+
// 覆盖:截断带尾部提示(上限/原长/memory_get id 指引,不再静默)/ 短正文无
|
|
3
|
+
// 提示 / 上限可配(injectContentMaxChars)/ 块预算随上限放大(调大上限不被
|
|
4
|
+
// 1500 旧闸卡死)/ _full_content 压缩注入路径不加提示 / 默认 300 与既有行为
|
|
5
|
+
// 对齐(ellipsis 仍在、全文不整段进入)。
|
|
6
|
+
import test from "node:test";
|
|
7
|
+
import assert from "node:assert/strict";
|
|
8
|
+
import { createStore } from "../src/store.js";
|
|
9
|
+
import { createService } from "../src/service.js";
|
|
10
|
+
import { createInjector } from "../src/inject.js";
|
|
11
|
+
import { createSettings } from "../src/settings.js";
|
|
12
|
+
|
|
13
|
+
function setup(over = {}) {
|
|
14
|
+
const store = createStore(":memory:");
|
|
15
|
+
const service = createService({ store, mirror: null, config: {} });
|
|
16
|
+
const settings = createSettings(store.db);
|
|
17
|
+
const contexts = [];
|
|
18
|
+
const ctx = {
|
|
19
|
+
systemPrompt: {
|
|
20
|
+
context(def) {
|
|
21
|
+
contexts.push(def);
|
|
22
|
+
return () => {};
|
|
23
|
+
}
|
|
24
|
+
}
|
|
25
|
+
};
|
|
26
|
+
const config = { maxInjectedItems: 3, importanceThreshold: 3, ...over };
|
|
27
|
+
const injector = createInjector(ctx, service, settings, config);
|
|
28
|
+
return { store, service, contexts, injector };
|
|
29
|
+
}
|
|
30
|
+
|
|
31
|
+
test("truncated entry carries a non-silent tail hint with limit/length/memory_get", () => {
|
|
32
|
+
const { contexts, service } = setup();
|
|
33
|
+
const long = "超长正文内容。".repeat(200); // ~1400 chars
|
|
34
|
+
const mem = service.saveWithDedupe({ type: "preference", title: "长记忆", content: long, importance: 5 }).memory;
|
|
35
|
+
const text = contexts[0].text({});
|
|
36
|
+
assert.ok(text.includes("…"), "ellipsis present");
|
|
37
|
+
assert.ok(text.includes(`memory_get "${mem.id}"`), "hint carries memory_get id");
|
|
38
|
+
assert.ok(text.includes("上限 300"), "hint carries the limit");
|
|
39
|
+
assert.ok(text.includes(`原 ${long.length} 字符`), "hint carries original length");
|
|
40
|
+
});
|
|
41
|
+
|
|
42
|
+
test("short content gets no hint", () => {
|
|
43
|
+
const { contexts, service } = setup();
|
|
44
|
+
service.saveWithDedupe({ type: "preference", title: "短记忆", content: "中文交流", importance: 5 });
|
|
45
|
+
const text = contexts[0].text({});
|
|
46
|
+
assert.ok(!text.includes("已截断"), "no truncation hint on short content");
|
|
47
|
+
});
|
|
48
|
+
|
|
49
|
+
test("injectContentMaxChars raises the cap and the hint reports the new limit", () => {
|
|
50
|
+
const { contexts, service } = setup({ injectContentMaxChars: 800 });
|
|
51
|
+
const long = "字".repeat(1000);
|
|
52
|
+
service.saveWithDedupe({ type: "preference", title: "长记忆", content: long, importance: 5 });
|
|
53
|
+
const text = contexts[0].text({});
|
|
54
|
+
assert.ok(text.includes("字".repeat(800)), "content injected up to the new cap");
|
|
55
|
+
assert.ok(!text.includes("字".repeat(801)), "content beyond the new cap is cut");
|
|
56
|
+
assert.ok(text.includes("上限 800"), "hint reports the configured limit");
|
|
57
|
+
});
|
|
58
|
+
|
|
59
|
+
test("block budget scales with the cap (raised cap is not defeated by the 1500 gate)", () => {
|
|
60
|
+
const { contexts, service } = setup({ injectContentMaxChars: 1000, maxInjectedItems: 2 });
|
|
61
|
+
const body = "块".repeat(900);
|
|
62
|
+
service.saveWithDedupe({ type: "preference", title: "甲", content: body, importance: 5 });
|
|
63
|
+
service.saveWithDedupe({ type: "preference", title: "乙", content: body, importance: 5 });
|
|
64
|
+
const text = contexts[0].text({});
|
|
65
|
+
assert.ok(text.includes("甲"), "first entry full");
|
|
66
|
+
assert.ok(text.includes("块".repeat(900)), "first entry body not collapsed by stale block gate");
|
|
67
|
+
assert.ok(text.includes("乙"), "second entry still rendered");
|
|
68
|
+
});
|
|
69
|
+
|
|
70
|
+
test("_full_content (sleep-compressed) path obeys the cap and says so (#266 评审)", () => {
|
|
71
|
+
const { store, contexts, service } = setup();
|
|
72
|
+
const mem = service.saveWithDedupe({ type: "preference", title: "压缩记忆", content: "占位", importance: 5 }).memory;
|
|
73
|
+
// 模拟 sleep 降级:content=压缩摘要(400 字符 > 默认 300 上限),原文入库 _full_content
|
|
74
|
+
const summary = "压缩摘要。".repeat(80);
|
|
75
|
+
store.db.prepare("UPDATE memories SET content=?, _full_content=? WHERE id=?")
|
|
76
|
+
.run(summary, "原始长文".repeat(100), mem.id);
|
|
77
|
+
const text = contexts[0].text({});
|
|
78
|
+
// 「降级过」不等于「天然够短」:原先这条早返回的唯一效果就是绕过上限,现在统一截断。
|
|
79
|
+
assert.ok(!text.includes(summary), "oversized summary is not injected whole");
|
|
80
|
+
assert.ok(text.includes("压缩摘要。".repeat(60)), "the part inside the cap is still there (300 chars)");
|
|
81
|
+
assert.ok(text.includes("已截断"), "the cut is reported, never silent");
|
|
82
|
+
});
|
|
83
|
+
|
|
84
|
+
test("default cap stays 300 (ellipsis regression of Bug6)", () => {
|
|
85
|
+
const { contexts, service } = setup();
|
|
86
|
+
const long = "x".repeat(2000);
|
|
87
|
+
service.saveWithDedupe({ type: "preference", title: "长记忆", content: long, importance: 5 });
|
|
88
|
+
const text = contexts[0].text({});
|
|
89
|
+
assert.ok(text.includes("x".repeat(300)));
|
|
90
|
+
assert.ok(!text.includes("x".repeat(301)));
|
|
91
|
+
});
|