sensemaking 0.24.5 → 0.24.7
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/dist/cjs/lib/watch-notification.d.cts +1 -0
- package/dist/cjs/lib/watch-notification.d.ts +1 -0
- package/dist/cjs/lib/watch-notification.js +15 -0
- package/dist/cjs/lib/watch-notification.js.map +1 -0
- package/dist/cjs/store/duckdb/native.d.cts +2 -2
- package/dist/cjs/store/duckdb/native.d.ts +2 -2
- package/dist/cjs/store/duckdb/native.js +1 -1
- package/dist/cjs/store/duckdb/native.js.map +1 -1
- package/dist/cjs/store/duckdb/open.js +33 -26
- package/dist/cjs/store/duckdb/open.js.map +1 -1
- package/dist/cjs/store/duckdb/vectors.d.cts +1 -0
- package/dist/cjs/store/duckdb/vectors.d.ts +1 -0
- package/dist/cjs/store/duckdb/vectors.js +55 -43
- package/dist/cjs/store/duckdb/vectors.js.map +1 -1
- package/dist/cjs/watch.js +8 -8
- package/dist/cjs/watch.js.map +1 -1
- package/dist/esm/lib/watch-notification.d.ts +1 -0
- package/dist/esm/lib/watch-notification.js +4 -0
- package/dist/esm/lib/watch-notification.js.map +1 -0
- package/dist/esm/store/duckdb/native.d.ts +2 -2
- package/dist/esm/store/duckdb/native.js +1 -1
- package/dist/esm/store/duckdb/native.js.map +1 -1
- package/dist/esm/store/duckdb/open.js +3 -1
- package/dist/esm/store/duckdb/open.js.map +1 -1
- package/dist/esm/store/duckdb/vectors.d.ts +1 -0
- package/dist/esm/store/duckdb/vectors.js +18 -14
- package/dist/esm/store/duckdb/vectors.js.map +1 -1
- package/dist/esm/watch.js +2 -2
- package/dist/esm/watch.js.map +1 -1
- package/package.json +2 -2
- package/skills/sense-setup/references/store-benchmarks.md +5 -5
|
@@ -0,0 +1 @@
|
|
|
1
|
+
export declare function shouldReconcileWatchNotification(filename: string | Buffer | null): boolean;
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
export declare function shouldReconcileWatchNotification(filename: string | Buffer | null): boolean;
|
|
@@ -0,0 +1,15 @@
|
|
|
1
|
+
"use strict";
|
|
2
|
+
Object.defineProperty(exports, "__esModule", {
|
|
3
|
+
value: true
|
|
4
|
+
});
|
|
5
|
+
Object.defineProperty(exports, "shouldReconcileWatchNotification", {
|
|
6
|
+
enumerable: true,
|
|
7
|
+
get: function() {
|
|
8
|
+
return shouldReconcileWatchNotification;
|
|
9
|
+
}
|
|
10
|
+
});
|
|
11
|
+
var _typests = require("../config/types.js");
|
|
12
|
+
function shouldReconcileWatchNotification(filename) {
|
|
13
|
+
return typeof filename !== 'string' || !filename.startsWith(_typests.STATE_DIR);
|
|
14
|
+
}
|
|
15
|
+
/* CJS INTEROP */ if (exports.__esModule && exports.default) { try { Object.defineProperty(exports.default, '__esModule', { value: true }); for (var key in exports) { exports.default[key] = exports[key]; } } catch (_) {}; module.exports = exports.default; }
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/lib/watch-notification.ts"],"sourcesContent":["import { STATE_DIR } from '../config/types.ts';\n\nexport function shouldReconcileWatchNotification(filename: string | Buffer | null): boolean {\n return typeof filename !== 'string' || !filename.startsWith(STATE_DIR);\n}\n"],"names":["shouldReconcileWatchNotification","filename","startsWith","STATE_DIR"],"mappings":";;;;+BAEgBA;;;eAAAA;;;uBAFU;AAEnB,SAASA,iCAAiCC,QAAgC;IAC/E,OAAO,OAAOA,aAAa,YAAY,CAACA,SAASC,UAAU,CAACC,kBAAS;AACvE"}
|
|
@@ -1,4 +1,4 @@
|
|
|
1
1
|
export declare const DUCKDB_PACKAGE = "@duckdb/node-api";
|
|
2
|
-
export declare const DUCKDB_VERSION = "1.5.5-r.
|
|
3
|
-
export declare const DUCKDB_INSTALL_SPEC = "@duckdb/node-api@1.5.5-r.
|
|
2
|
+
export declare const DUCKDB_VERSION = "1.5.5-r.5";
|
|
3
|
+
export declare const DUCKDB_INSTALL_SPEC = "@duckdb/node-api@1.5.5-r.5";
|
|
4
4
|
export declare function duckdbApi(): Promise<typeof import('@duckdb/node-api')>;
|
|
@@ -1,4 +1,4 @@
|
|
|
1
1
|
export declare const DUCKDB_PACKAGE = "@duckdb/node-api";
|
|
2
|
-
export declare const DUCKDB_VERSION = "1.5.5-r.
|
|
3
|
-
export declare const DUCKDB_INSTALL_SPEC = "@duckdb/node-api@1.5.5-r.
|
|
2
|
+
export declare const DUCKDB_VERSION = "1.5.5-r.5";
|
|
3
|
+
export declare const DUCKDB_INSTALL_SPEC = "@duckdb/node-api@1.5.5-r.5";
|
|
4
4
|
export declare function duckdbApi(): Promise<typeof import('@duckdb/node-api')>;
|
|
@@ -24,7 +24,7 @@ _export(exports, {
|
|
|
24
24
|
});
|
|
25
25
|
var _nativets = require("../native.js");
|
|
26
26
|
var DUCKDB_PACKAGE = '@duckdb/node-api';
|
|
27
|
-
var DUCKDB_VERSION = '1.5.5-r.
|
|
27
|
+
var DUCKDB_VERSION = '1.5.5-r.5';
|
|
28
28
|
var DUCKDB_INSTALL_SPEC = "".concat(DUCKDB_PACKAGE, "@").concat(DUCKDB_VERSION);
|
|
29
29
|
var DUCKDB = {
|
|
30
30
|
store: 'duckdb',
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/store/duckdb/native.ts"],"sourcesContent":["import { loadOrInstall, type NativeDescriptor, packageNodeModules } from '../native.ts';\n\nexport const DUCKDB_PACKAGE = '@duckdb/node-api';\nexport const DUCKDB_VERSION = '1.5.5-r.
|
|
1
|
+
{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/store/duckdb/native.ts"],"sourcesContent":["import { loadOrInstall, type NativeDescriptor, packageNodeModules } from '../native.ts';\n\nexport const DUCKDB_PACKAGE = '@duckdb/node-api';\nexport const DUCKDB_VERSION = '1.5.5-r.5';\nexport const DUCKDB_INSTALL_SPEC = `${DUCKDB_PACKAGE}@${DUCKDB_VERSION}`;\n\nconst DUCKDB: NativeDescriptor = {\n store: 'duckdb',\n pkg: DUCKDB_PACKAGE,\n sizeHint: '~110MB',\n version: DUCKDB_VERSION,\n installSpec: DUCKDB_INSTALL_SPEC,\n};\n\nlet duckdbApiPromise: Promise<typeof import('@duckdb/node-api')> | undefined;\n\n// The one accessor every consumer of @duckdb/node-api must go through: after a fresh install in this process, Node keeps returning\n// the pre-install \"not found\" miss to any new bare-specifier lookup of the same path, so caching the resolved module here means only the first caller ever resolves it.\nexport function duckdbApi(): Promise<typeof import('@duckdb/node-api')> {\n if (!duckdbApiPromise) duckdbApiPromise = loadOrInstall<typeof import('@duckdb/node-api')>(DUCKDB, packageNodeModules());\n return duckdbApiPromise;\n}\n"],"names":["DUCKDB_INSTALL_SPEC","DUCKDB_PACKAGE","DUCKDB_VERSION","duckdbApi","DUCKDB","store","pkg","sizeHint","version","installSpec","duckdbApiPromise","loadOrInstall","packageNodeModules"],"mappings":";;;;;;;;;;;QAIaA;eAAAA;;QAFAC;eAAAA;;QACAC;eAAAA;;QAeGC;eAAAA;;;wBAlByD;AAElE,IAAMF,iBAAiB;AACvB,IAAMC,iBAAiB;AACvB,IAAMF,sBAAsB,AAAC,GAAoBE,OAAlBD,gBAAe,KAAkB,OAAfC;AAExD,IAAME,SAA2B;IAC/BC,OAAO;IACPC,KAAKL;IACLM,UAAU;IACVC,SAASN;IACTO,aAAaT;AACf;AAEA,IAAIU;AAIG,SAASP;IACd,IAAI,CAACO,kBAAkBA,mBAAmBC,IAAAA,uBAAa,EAAoCP,QAAQQ,IAAAA,4BAAkB;IACrH,OAAOF;AACT"}
|
|
@@ -33,6 +33,7 @@ var _nativets = require("./native.js");
|
|
|
33
33
|
var _reconcilets = require("./reconcile.js");
|
|
34
34
|
var _sqlfunctionsts = require("./sql-functions.js");
|
|
35
35
|
var _storets = require("./store.js");
|
|
36
|
+
var _vectorsts = require("./vectors.js");
|
|
36
37
|
function asyncGeneratorStep(gen, resolve, reject, _next, _throw, key, arg) {
|
|
37
38
|
try {
|
|
38
39
|
var info = gen[key](arg);
|
|
@@ -168,7 +169,7 @@ var DB_FILENAME = 'cache.duckdb';
|
|
|
168
169
|
var SCHEMA_VERSION = '4';
|
|
169
170
|
// `content` is a plain table (not FTS-virtual): the fts index is built over it lazily, only when a lexical query runs (lexical.ts), and read directly for contains() verification either way.
|
|
170
171
|
// No tokenizer resolution: this store always uses the fts extension's default (porter) stemmer.
|
|
171
|
-
function ensureSchema(
|
|
172
|
+
function ensureSchema(handle, conn, cfg) {
|
|
172
173
|
return _async_to_generator(function() {
|
|
173
174
|
var _iteratorNormalCompletion, _didIteratorError, _iteratorError, _iterator, _step, feature, err;
|
|
174
175
|
return _ts_generator(this, function(_state) {
|
|
@@ -203,21 +204,21 @@ function ensureSchema(_handle, conn, cfg) {
|
|
|
203
204
|
case 5:
|
|
204
205
|
_state.trys.push([
|
|
205
206
|
5,
|
|
206
|
-
12,
|
|
207
207
|
13,
|
|
208
|
-
14
|
|
208
|
+
14,
|
|
209
|
+
15
|
|
209
210
|
]);
|
|
210
211
|
_iterator = (0, _indexts1.activeFeatures)(cfg)[Symbol.iterator]();
|
|
211
212
|
_state.label = 6;
|
|
212
213
|
case 6:
|
|
213
214
|
if (!!(_iteratorNormalCompletion = (_step = _iterator.next()).done)) return [
|
|
214
215
|
3,
|
|
215
|
-
|
|
216
|
+
12
|
|
216
217
|
];
|
|
217
218
|
feature = _step.value;
|
|
218
219
|
if (!(feature.name === 'embed')) return [
|
|
219
220
|
3,
|
|
220
|
-
|
|
221
|
+
9
|
|
221
222
|
];
|
|
222
223
|
return [
|
|
223
224
|
4,
|
|
@@ -226,37 +227,43 @@ function ensureSchema(_handle, conn, cfg) {
|
|
|
226
227
|
case 7:
|
|
227
228
|
_state.sent();
|
|
228
229
|
return [
|
|
229
|
-
|
|
230
|
-
|
|
230
|
+
4,
|
|
231
|
+
(0, _vectorsts.createVectorWriteStage)(handle.duckdb, _typests.STORE_DIMS)
|
|
231
232
|
];
|
|
232
233
|
case 8:
|
|
234
|
+
_state.sent();
|
|
235
|
+
return [
|
|
236
|
+
3,
|
|
237
|
+
11
|
|
238
|
+
];
|
|
239
|
+
case 9:
|
|
233
240
|
return [
|
|
234
241
|
4,
|
|
235
242
|
feature.schema(conn)
|
|
236
243
|
];
|
|
237
|
-
case 9:
|
|
238
|
-
_state.sent();
|
|
239
|
-
_state.label = 10;
|
|
240
244
|
case 10:
|
|
245
|
+
_state.sent();
|
|
246
|
+
_state.label = 11;
|
|
247
|
+
case 11:
|
|
241
248
|
_iteratorNormalCompletion = true;
|
|
242
249
|
return [
|
|
243
250
|
3,
|
|
244
251
|
6
|
|
245
252
|
];
|
|
246
|
-
case
|
|
253
|
+
case 12:
|
|
247
254
|
return [
|
|
248
255
|
3,
|
|
249
|
-
|
|
256
|
+
15
|
|
250
257
|
];
|
|
251
|
-
case
|
|
258
|
+
case 13:
|
|
252
259
|
err = _state.sent();
|
|
253
260
|
_didIteratorError = true;
|
|
254
261
|
_iteratorError = err;
|
|
255
262
|
return [
|
|
256
263
|
3,
|
|
257
|
-
|
|
264
|
+
15
|
|
258
265
|
];
|
|
259
|
-
case
|
|
266
|
+
case 14:
|
|
260
267
|
try {
|
|
261
268
|
if (!_iteratorNormalCompletion && _iterator.return != null) {
|
|
262
269
|
_iterator.return();
|
|
@@ -269,41 +276,41 @@ function ensureSchema(_handle, conn, cfg) {
|
|
|
269
276
|
return [
|
|
270
277
|
7
|
|
271
278
|
];
|
|
272
|
-
case
|
|
279
|
+
case 15:
|
|
273
280
|
return [
|
|
274
281
|
4,
|
|
275
282
|
(0, _sharedts.getMeta)(conn, 'schema_version')
|
|
276
283
|
];
|
|
277
|
-
case
|
|
284
|
+
case 16:
|
|
278
285
|
if (!(_state.sent() === null)) return [
|
|
279
286
|
3,
|
|
280
|
-
|
|
287
|
+
18
|
|
281
288
|
];
|
|
282
289
|
return [
|
|
283
290
|
4,
|
|
284
291
|
(0, _sharedts.setMeta)(conn, 'schema_version', SCHEMA_VERSION)
|
|
285
292
|
];
|
|
286
|
-
case 16:
|
|
287
|
-
_state.sent();
|
|
288
|
-
_state.label = 17;
|
|
289
293
|
case 17:
|
|
294
|
+
_state.sent();
|
|
295
|
+
_state.label = 18;
|
|
296
|
+
case 18:
|
|
290
297
|
return [
|
|
291
298
|
4,
|
|
292
299
|
(0, _sharedts.getMeta)(conn, 'features')
|
|
293
300
|
];
|
|
294
|
-
case
|
|
301
|
+
case 19:
|
|
295
302
|
if (!(_state.sent() === null)) return [
|
|
296
303
|
3,
|
|
297
|
-
|
|
304
|
+
21
|
|
298
305
|
];
|
|
299
306
|
return [
|
|
300
307
|
4,
|
|
301
308
|
(0, _sharedts.setMeta)(conn, 'features', (0, _indexts.featureSignature)(cfg, _indexts1.FEATURES))
|
|
302
309
|
];
|
|
303
|
-
case 19:
|
|
304
|
-
_state.sent();
|
|
305
|
-
_state.label = 20;
|
|
306
310
|
case 20:
|
|
311
|
+
_state.sent();
|
|
312
|
+
_state.label = 21;
|
|
313
|
+
case 21:
|
|
307
314
|
return [
|
|
308
315
|
2
|
|
309
316
|
];
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/store/duckdb/open.ts"],"sourcesContent":["import type { DuckDBConnection, DuckDBInstance } from '@duckdb/node-api';\nimport type { Config, ResolvedConfig } from '../../config/index.ts';\nimport { featureSignature } from '../../config/index.ts';\nimport { STORE_DIMS } from '../../embed/types.ts';\nimport { SenseError } from '../../errors.ts';\nimport { activeFeatures, FEATURES } from '../../features/index.ts';\nimport type { OpenResult } from '../open.ts';\nimport { openWithDialect } from '../open.ts';\nimport { getMeta, setMeta } from '../shared.ts';\nimport type { Connection, OpenDialect } from '../types.ts';\nimport { createConnection } from './connection.ts';\nimport { DUCKDB_PACKAGE, duckdbApi } from './native.ts';\nimport { duckdbDialect } from './reconcile.ts';\nimport { registerFunctions } from './sql-functions.ts';\nimport { createStore } from './store.ts';\n\nexport const DB_FILENAME = 'cache.duckdb';\n// Independent of sqlite's SCHEMA_VERSION -- the two stores' cache shapes evolve separately (VARIANT frontmatter columns vs untyped).\n// The store name already joins the feature signature, so switching a config's `store` key rebuilds rather than reusing the other engine's cache.\nexport const SCHEMA_VERSION = '4';\n\nexport type { OpenResult };\n\n// The connection plus this store's own native handles (types.ts's OpenDialect<Handle>): the\n// instance owns the WAL and must outlive the connection borrowed from it.\ninterface DuckdbHandle {\n instance: DuckDBInstance;\n duckdb: DuckDBConnection;\n}\n\n// `content` is a plain table (not FTS-virtual): the fts index is built over it lazily, only when a lexical query runs (lexical.ts), and read directly for contains() verification either way.\n// No tokenizer resolution: this store always uses the fts extension's default (porter) stemmer.\nasync function ensureSchema(
|
|
1
|
+
{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/store/duckdb/open.ts"],"sourcesContent":["import type { DuckDBConnection, DuckDBInstance } from '@duckdb/node-api';\nimport type { Config, ResolvedConfig } from '../../config/index.ts';\nimport { featureSignature } from '../../config/index.ts';\nimport { STORE_DIMS } from '../../embed/types.ts';\nimport { SenseError } from '../../errors.ts';\nimport { activeFeatures, FEATURES } from '../../features/index.ts';\nimport type { OpenResult } from '../open.ts';\nimport { openWithDialect } from '../open.ts';\nimport { getMeta, setMeta } from '../shared.ts';\nimport type { Connection, OpenDialect } from '../types.ts';\nimport { createConnection } from './connection.ts';\nimport { DUCKDB_PACKAGE, duckdbApi } from './native.ts';\nimport { duckdbDialect } from './reconcile.ts';\nimport { registerFunctions } from './sql-functions.ts';\nimport { createStore } from './store.ts';\nimport { createVectorWriteStage } from './vectors.ts';\n\nexport const DB_FILENAME = 'cache.duckdb';\n// Independent of sqlite's SCHEMA_VERSION -- the two stores' cache shapes evolve separately (VARIANT frontmatter columns vs untyped).\n// The store name already joins the feature signature, so switching a config's `store` key rebuilds rather than reusing the other engine's cache.\nexport const SCHEMA_VERSION = '4';\n\nexport type { OpenResult };\n\n// The connection plus this store's own native handles (types.ts's OpenDialect<Handle>): the\n// instance owns the WAL and must outlive the connection borrowed from it.\ninterface DuckdbHandle {\n instance: DuckDBInstance;\n duckdb: DuckDBConnection;\n}\n\n// `content` is a plain table (not FTS-virtual): the fts index is built over it lazily, only when a lexical query runs (lexical.ts), and read directly for contains() verification either way.\n// No tokenizer resolution: this store always uses the fts extension's default (porter) stemmer.\nasync function ensureSchema(handle: DuckdbHandle, conn: Connection, cfg: Config): Promise<void> {\n await conn.exec(`CREATE TABLE IF NOT EXISTS frontmatter (\"path\" TEXT PRIMARY KEY, \"_mtime\" DOUBLE, \"_ctime\" DOUBLE, \"_size\" INTEGER, \"_parse_error\" TEXT)`);\n await conn.exec(`CREATE TABLE IF NOT EXISTS content (\"path\" TEXT PRIMARY KEY, title TEXT, summary TEXT, text TEXT)`);\n await conn.exec(`CREATE TABLE IF NOT EXISTS preset_files (\"path\" TEXT, preset TEXT, PRIMARY KEY (\"path\", preset))`);\n await conn.exec('CREATE INDEX IF NOT EXISTS preset_files_preset ON preset_files(preset)');\n for (const feature of activeFeatures(cfg)) {\n // Native FLOAT[STORE_DIMS] instead of the embed feature's engine-neutral BLOB+scale DDL (vectors.ts). `scale` is kept, unused,\n // so the feature's shared reconcile-time INSERT/DELETE (features/embed.ts) names a column that exists on both stores.\n if (feature.name === 'embed') {\n await conn.exec(`CREATE TABLE IF NOT EXISTS embeddings (\"path\" TEXT, chunk INTEGER, start_line INTEGER, end_line INTEGER, scale REAL, vector FLOAT[${STORE_DIMS}], PRIMARY KEY (\"path\", chunk))`);\n await createVectorWriteStage(handle.duckdb, STORE_DIMS);\n continue;\n }\n await feature.schema(conn);\n }\n if ((await getMeta(conn, 'schema_version')) === null) await setMeta(conn, 'schema_version', SCHEMA_VERSION);\n if ((await getMeta(conn, 'features')) === null) await setMeta(conn, 'features', featureSignature(cfg, FEATURES));\n}\n\n// Order matters: the connection must be gone before the instance closes the WAL.\nasync function close(handle: DuckdbHandle): Promise<void> {\n handle.duckdb.disconnectSync();\n handle.instance.closeSync();\n}\n\nasync function connect(dbPath: string, _cfg: ResolvedConfig): Promise<{ handle: DuckdbHandle; conn: Connection }> {\n // Dynamic, not a top-level import: sqlite trees must never attempt to resolve this optional peer dependency, so nothing imports\n // it as a value until a duckdb tree opens (types-only imports are erased). Installed on first use if missing, shared with sql-functions.ts and vectors.ts via native.ts's duckdbApi.\n let DuckDBInstance: typeof import('@duckdb/node-api').DuckDBInstance;\n try {\n ({ DuckDBInstance } = await duckdbApi());\n } catch (err) {\n if (err instanceof SenseError) throw err;\n throw new SenseError('STORE_DEPENDENCY_MISSING', `store \"duckdb\" needs the ${DUCKDB_PACKAGE} package (${(err as Error).message})`);\n }\n\n let duckdb: DuckDBConnection;\n let instance: DuckDBInstance | undefined;\n try {\n // On-disk files default to an older storage format for cross-version compatibility, which rejects VARIANT columns (\"VARIANT\n // columns are not supported in storage versions prior to v1.5.0\"); this store's dynamic frontmatter columns need VARIANT (reconcile.ts), so the floor is pinned explicitly.\n instance = await DuckDBInstance.create(dbPath, { storage_compatibility_version: 'v1.5.0' });\n duckdb = await instance.connect();\n } catch (err) {\n // create() may have succeeded before connect() failed: close it, or its WAL stays open.\n instance?.closeSync();\n throw new SenseError('STORE_DEPENDENCY_MISSING', `store \"duckdb\" failed to open ${dbPath}: ${(err as Error).message}`);\n }\n // A throw below would leak the open instance, whose WAL then locks the .duckdb file undeletable on Windows.\n try {\n await registerFunctions(duckdb);\n const conn = createConnection(duckdb);\n return { handle: { instance, duckdb }, conn };\n } catch (err) {\n await close({ instance, duckdb });\n throw err;\n }\n}\n\n// This store's dialect (types.ts's OpenDialect) for the shared orchestration in store/open.ts.\nexport const duckdbOpenDialect: OpenDialect<DuckdbHandle> = {\n filename: DB_FILENAME,\n schemaVersion: SCHEMA_VERSION,\n reconcileDialect: duckdbDialect,\n connect,\n close,\n // duckdb's file lock spans the connection's life, and each platform words the refusal its own\n // way: posix \"Could not set lock on file ... Conflicting lock is held in <exe> (PID n)\", Windows\n // \"Cannot open file ... being used by another process\". Both are the same condition.\n isLocked: (err) => /Could not set lock on file|being used by another process/.test(err.message),\n ensureSchema,\n createStore: (handle, conn) => createStore(handle.instance, handle.duckdb, conn),\n};\n\nexport async function openDuckdb(cfg: ResolvedConfig): Promise<OpenResult> {\n return openWithDialect(cfg, duckdbOpenDialect);\n}\n"],"names":["DB_FILENAME","SCHEMA_VERSION","duckdbOpenDialect","openDuckdb","ensureSchema","handle","conn","cfg","feature","exec","activeFeatures","name","STORE_DIMS","createVectorWriteStage","duckdb","schema","getMeta","setMeta","featureSignature","FEATURES","close","disconnectSync","instance","closeSync","connect","dbPath","_cfg","DuckDBInstance","err","duckdbApi","SenseError","DUCKDB_PACKAGE","message","create","storage_compatibility_version","registerFunctions","createConnection","filename","schemaVersion","reconcileDialect","duckdbDialect","isLocked","test","createStore","openWithDialect"],"mappings":";;;;;;;;;;;QAiBaA;eAAAA;;QAGAC;eAAAA;;QAyEAC;eAAAA;;QAcSC;eAAAA;;;uBAzGW;uBACN;wBACA;wBACc;sBAET;wBACC;4BAEA;wBACS;2BACZ;8BACI;uBACN;yBACW;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAEhC,IAAMH,cAAc;AAGpB,IAAMC,iBAAiB;AAW9B,8LAA8L;AAC9L,gGAAgG;AAChG,SAAeG,aAAaC,MAAoB,EAAEC,IAAgB,EAAEC,GAAW;;YAKxE,2BAAA,mBAAA,gBAAA,WAAA,OAAMC;;;;oBAJX;;wBAAMF,KAAKG,IAAI,CAAC;;;oBAAhB;oBACA;;wBAAMH,KAAKG,IAAI,CAAC;;;oBAAhB;oBACA;;wBAAMH,KAAKG,IAAI,CAAC;;;oBAAhB;oBACA;;wBAAMH,KAAKG,IAAI,CAAC;;;oBAAhB;oBACK,kCAAA,2BAAA;;;;;;;;;oBAAA,YAAiBC,IAAAA,wBAAc,EAACH;;;2BAAhC,6BAAA,QAAA;;;;oBAAMC,UAAN;yBAGCA,CAAAA,QAAQG,IAAI,KAAK,OAAM,GAAvBH;;;;oBACF;;wBAAMF,KAAKG,IAAI,CAAC,AAAC,qIAA+I,OAAXG,mBAAU,EAAC;;;oBAAhK;oBACA;;wBAAMC,IAAAA,iCAAsB,EAACR,OAAOS,MAAM,EAAEF,mBAAU;;;oBAAtD;oBACA;;;;;oBAEF;;wBAAMJ,QAAQO,MAAM,CAACT;;;oBAArB;;;oBARG;;;;;;;;;;;;oBAAA;oBAAA;;;;;;;6BAAA,6BAAA;4BAAA;;;4BAAA;kCAAA;;;;;;;oBAUA;;wBAAMU,IAAAA,iBAAO,EAACV,MAAM;;;yBAArB,CAAA,AAAC,kBAA2C,IAAG,GAA/C;;;;oBAAkD;;wBAAMW,IAAAA,iBAAO,EAACX,MAAM,kBAAkBL;;;oBAAtC;;;oBACjD;;wBAAMe,IAAAA,iBAAO,EAACV,MAAM;;;yBAArB,CAAA,AAAC,kBAAqC,IAAG,GAAzC;;;;oBAA4C;;wBAAMW,IAAAA,iBAAO,EAACX,MAAM,YAAYY,IAAAA,yBAAgB,EAACX,KAAKY,kBAAQ;;;oBAA9D;;;;;;;;IAClD;;AAEA,iFAAiF;AACjF,SAAeC,MAAMf,MAAoB;;;YACvCA,OAAOS,MAAM,CAACO,cAAc;YAC5BhB,OAAOiB,QAAQ,CAACC,SAAS;;;;;IAC3B;;AAEA,SAAeC,QAAQC,MAAc,EAAEC,IAAoB;;YAGrDC,gBAGKC,KAKLd,QACAQ,UAMKM,MAQDtB,MAECsB;;;;;;;;;;oBAvBe;;wBAAMC,IAAAA,mBAAS;;;oBAAlCF,iBAAmB,cAAnBA;;;;;;oBACIC;oBACP,IAAIA,AAAG,YAAHA,KAAeE,oBAAU,GAAE,MAAMF;oBACrC,MAAM,IAAIE,oBAAU,CAAC,4BAA4B,AAAC,4BAAsD,OAA3BC,wBAAc,EAAC,cAAmC,OAAvB,AAACH,IAAcI,OAAO,EAAC;;;;;;;;oBAQpH;;wBAAML,eAAeM,MAAM,CAACR,QAAQ;4BAAES,+BAA+B;wBAAS;;;oBAFzF,4HAA4H;oBAC5H,4KAA4K;oBAC5KZ,WAAW;oBACF;;wBAAMA,SAASE,OAAO;;;oBAA/BV,SAAS;;;;;;oBACFc;oBACP,wFAAwF;oBACxFN,qBAAAA,+BAAAA,SAAUC,SAAS;oBACnB,MAAM,IAAIO,oBAAU,CAAC,4BAA4B,AAAC,iCAA2C,OAAXL,QAAO,MAA2B,OAAvB,AAACG,KAAcI,OAAO;;;;;;;;oBAInH;;wBAAMG,IAAAA,iCAAiB,EAACrB;;;oBAAxB;oBACMR,OAAO8B,IAAAA,8BAAgB,EAACtB;oBAC9B;;wBAAO;4BAAET,QAAQ;gCAAEiB,UAAAA;gCAAUR,QAAAA;4BAAO;4BAAGR,MAAAA;wBAAK;;;oBACrCsB;oBACP;;wBAAMR,MAAM;4BAAEE,UAAAA;4BAAUR,QAAAA;wBAAO;;;oBAA/B;oBACA,MAAMc;;;;;;;IAEV;;AAGO,IAAM1B,oBAA+C;IAC1DmC,UAAUrC;IACVsC,eAAerC;IACfsC,kBAAkBC,0BAAa;IAC/BhB,SAAAA;IACAJ,OAAAA;IACA,8FAA8F;IAC9F,iGAAiG;IACjG,qFAAqF;IACrFqB,UAAU,SAAVA,SAAWb;eAAQ,2DAA2Dc,IAAI,CAACd,IAAII,OAAO;;IAC9F5B,cAAAA;IACAuC,aAAa,SAAbA,YAActC,QAAQC;eAASqC,IAAAA,oBAAW,EAACtC,OAAOiB,QAAQ,EAAEjB,OAAOS,MAAM,EAAER;;AAC7E;AAEO,SAAeH,WAAWI,GAAmB;;;YAClD;;gBAAOqC,IAAAA,uBAAe,EAACrC,KAAKL;;;IAC9B"}
|
|
@@ -1,5 +1,6 @@
|
|
|
1
1
|
import type { DuckDBConnection } from '@duckdb/node-api';
|
|
2
2
|
import type { Connection, VectorCandidate, VectorSimilar, VectorWriteRow } from '../types.js';
|
|
3
|
+
export declare function createVectorWriteStage(duckdb: DuckDBConnection, dims: number): Promise<void>;
|
|
3
4
|
export declare function writeVectorBatch(duckdb: DuckDBConnection, conn: Connection, dims: number, rows: VectorWriteRow[]): Promise<void>;
|
|
4
5
|
export declare function scanCandidates(duckdb: DuckDBConnection, qv: Float32Array, dims: number, fetch: number, allowed?: Set<string>): Promise<VectorCandidate[]>;
|
|
5
6
|
export declare function scanSimilar(duckdb: DuckDBConnection, conn: Connection, dims: number, path: string, opts: {
|
|
@@ -1,5 +1,6 @@
|
|
|
1
1
|
import type { DuckDBConnection } from '@duckdb/node-api';
|
|
2
2
|
import type { Connection, VectorCandidate, VectorSimilar, VectorWriteRow } from '../types.js';
|
|
3
|
+
export declare function createVectorWriteStage(duckdb: DuckDBConnection, dims: number): Promise<void>;
|
|
3
4
|
export declare function writeVectorBatch(duckdb: DuckDBConnection, conn: Connection, dims: number, rows: VectorWriteRow[]): Promise<void>;
|
|
4
5
|
export declare function scanCandidates(duckdb: DuckDBConnection, qv: Float32Array, dims: number, fetch: number, allowed?: Set<string>): Promise<VectorCandidate[]>;
|
|
5
6
|
export declare function scanSimilar(duckdb: DuckDBConnection, conn: Connection, dims: number, path: string, opts: {
|
|
@@ -9,6 +9,9 @@ function _export(target, all) {
|
|
|
9
9
|
});
|
|
10
10
|
}
|
|
11
11
|
_export(exports, {
|
|
12
|
+
get createVectorWriteStage () {
|
|
13
|
+
return createVectorWriteStage;
|
|
14
|
+
},
|
|
12
15
|
get scanCandidates () {
|
|
13
16
|
return scanCandidates;
|
|
14
17
|
},
|
|
@@ -21,7 +24,6 @@ _export(exports, {
|
|
|
21
24
|
});
|
|
22
25
|
var _transactionts = require("../transaction.js");
|
|
23
26
|
var _vectorsts = require("../vectors.js");
|
|
24
|
-
var _batchts = require("./batch.js");
|
|
25
27
|
var _nativets = require("./native.js");
|
|
26
28
|
function _array_like_to_array(arr, len) {
|
|
27
29
|
if (len == null || len > arr.length) len = arr.length;
|
|
@@ -182,6 +184,25 @@ function _unsupported_iterable_to_array(o, minLen) {
|
|
|
182
184
|
// A bind position whose type is left to auto-inference (safe for plain strings/numbers; only
|
|
183
185
|
// the vector ARRAY positions below need an explicit type -- see writeVectorBatch's comment).
|
|
184
186
|
var untyped = undefined;
|
|
187
|
+
var VECTOR_WRITE_STAGE = '_sense_vector_write_stage';
|
|
188
|
+
function createVectorWriteStage(duckdb, dims) {
|
|
189
|
+
return _async_to_generator(function() {
|
|
190
|
+
return _ts_generator(this, function(_state) {
|
|
191
|
+
switch(_state.label){
|
|
192
|
+
case 0:
|
|
193
|
+
return [
|
|
194
|
+
4,
|
|
195
|
+
duckdb.run("CREATE TEMP TABLE ".concat(VECTOR_WRITE_STAGE, ' ("path" TEXT, chunk INTEGER, vector FLOAT[').concat(dims, "])"))
|
|
196
|
+
];
|
|
197
|
+
case 1:
|
|
198
|
+
_state.sent();
|
|
199
|
+
return [
|
|
200
|
+
2
|
|
201
|
+
];
|
|
202
|
+
}
|
|
203
|
+
});
|
|
204
|
+
})();
|
|
205
|
+
}
|
|
185
206
|
// The DDL-fixed array width can exceed a vector's actual length (a hypothetical model sliced under the column's width);
|
|
186
207
|
// zero-padding leaves cosine scores unchanged since the added dimensions contribute nothing to either vector's dot product or norm.
|
|
187
208
|
function padded(values, dims) {
|
|
@@ -206,7 +227,7 @@ function dequantize(row, dims) {
|
|
|
206
227
|
}
|
|
207
228
|
function writeVectorBatch(duckdb, conn, dims, rows) {
|
|
208
229
|
return _async_to_generator(function() {
|
|
209
|
-
var _ref,
|
|
230
|
+
var _ref, ARRAY, FLOAT, vectorType;
|
|
210
231
|
return _ts_generator(this, function(_state) {
|
|
211
232
|
switch(_state.label){
|
|
212
233
|
case 0:
|
|
@@ -218,71 +239,62 @@ function writeVectorBatch(duckdb, conn, dims, rows) {
|
|
|
218
239
|
(0, _nativets.duckdbApi)()
|
|
219
240
|
];
|
|
220
241
|
case 1:
|
|
221
|
-
_ref = _state.sent(),
|
|
222
|
-
|
|
223
|
-
if (!rewritten) throw new Error('duckdb: writeVectorBatch SQL shape not recognized by rewriteUpdate');
|
|
242
|
+
_ref = _state.sent(), ARRAY = _ref.ARRAY, FLOAT = _ref.FLOAT;
|
|
243
|
+
vectorType = ARRAY(FLOAT, dims);
|
|
224
244
|
return [
|
|
225
245
|
4,
|
|
226
246
|
(0, _transactionts.withTransaction)(conn, function() {
|
|
227
247
|
return _async_to_generator(function() {
|
|
228
|
-
var
|
|
248
|
+
var appender, _iteratorNormalCompletion, _didIteratorError, _iteratorError, _iterator, _step, row;
|
|
229
249
|
return _ts_generator(this, function(_state) {
|
|
230
250
|
switch(_state.label){
|
|
231
251
|
case 0:
|
|
232
252
|
return [
|
|
233
253
|
4,
|
|
234
|
-
duckdb.
|
|
254
|
+
duckdb.run("DELETE FROM ".concat(VECTOR_WRITE_STAGE))
|
|
235
255
|
];
|
|
236
256
|
case 1:
|
|
237
|
-
|
|
238
|
-
|
|
239
|
-
case 2:
|
|
240
|
-
_state.trys.push([
|
|
241
|
-
2,
|
|
242
|
-
,
|
|
257
|
+
_state.sent();
|
|
258
|
+
return [
|
|
243
259
|
4,
|
|
244
|
-
|
|
245
|
-
]
|
|
246
|
-
|
|
247
|
-
|
|
248
|
-
_iteratorNormalCompletion = true, _didIteratorError = false, _iteratorError = undefined;
|
|
260
|
+
duckdb.createAppender(VECTOR_WRITE_STAGE)
|
|
261
|
+
];
|
|
262
|
+
case 2:
|
|
263
|
+
appender = _state.sent();
|
|
249
264
|
try {
|
|
250
|
-
|
|
251
|
-
row = _step.value;
|
|
252
|
-
values.push(arrayValue(dequantize(row, dims)), row.path, row.chunk);
|
|
253
|
-
types.push(ARRAY(DOUBLE, dims), untyped, untyped);
|
|
254
|
-
}
|
|
255
|
-
} catch (err) {
|
|
256
|
-
_didIteratorError = true;
|
|
257
|
-
_iteratorError = err;
|
|
258
|
-
} finally{
|
|
265
|
+
_iteratorNormalCompletion = true, _didIteratorError = false, _iteratorError = undefined;
|
|
259
266
|
try {
|
|
260
|
-
|
|
261
|
-
|
|
267
|
+
for(_iterator = rows[Symbol.iterator](); !(_iteratorNormalCompletion = (_step = _iterator.next()).done); _iteratorNormalCompletion = true){
|
|
268
|
+
row = _step.value;
|
|
269
|
+
appender.appendVarchar(row.path);
|
|
270
|
+
appender.appendInteger(row.chunk);
|
|
271
|
+
appender.appendArray(dequantize(row, dims), vectorType);
|
|
272
|
+
appender.endRow();
|
|
262
273
|
}
|
|
274
|
+
} catch (err) {
|
|
275
|
+
_didIteratorError = true;
|
|
276
|
+
_iteratorError = err;
|
|
263
277
|
} finally{
|
|
264
|
-
|
|
265
|
-
|
|
278
|
+
try {
|
|
279
|
+
if (!_iteratorNormalCompletion && _iterator.return != null) {
|
|
280
|
+
_iterator.return();
|
|
281
|
+
}
|
|
282
|
+
} finally{
|
|
283
|
+
if (_didIteratorError) {
|
|
284
|
+
throw _iteratorError;
|
|
285
|
+
}
|
|
266
286
|
}
|
|
267
287
|
}
|
|
288
|
+
appender.flushSync();
|
|
289
|
+
} finally{
|
|
290
|
+
appender.closeSync();
|
|
268
291
|
}
|
|
269
|
-
stmt.bind(values, types);
|
|
270
292
|
return [
|
|
271
293
|
4,
|
|
272
|
-
|
|
294
|
+
duckdb.run("UPDATE embeddings SET vector = staged.vector FROM ".concat(VECTOR_WRITE_STAGE, ' AS staged WHERE embeddings."path" = staged."path" AND embeddings.chunk = staged.chunk'))
|
|
273
295
|
];
|
|
274
296
|
case 3:
|
|
275
297
|
_state.sent();
|
|
276
|
-
return [
|
|
277
|
-
3,
|
|
278
|
-
5
|
|
279
|
-
];
|
|
280
|
-
case 4:
|
|
281
|
-
stmt.destroySync();
|
|
282
|
-
return [
|
|
283
|
-
7
|
|
284
|
-
];
|
|
285
|
-
case 5:
|
|
286
298
|
return [
|
|
287
299
|
2
|
|
288
300
|
];
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/store/duckdb/vectors.ts"],"sourcesContent":["import type { DuckDBConnection, DuckDBType, DuckDBValue } from '@duckdb/node-api';\nimport { withTransaction } from '../transaction.ts';\nimport type { Connection, VectorCandidate, VectorSimilar, VectorWriteRow } from '../types.ts';\nimport { asCosine, sampleEvenly } from '../vectors.ts';\nimport { rewriteUpdate } from './batch.ts';\nimport { duckdbApi } from './native.ts';\n\n// This store keeps vectors as native FLOAT[dims] arrays (dims fixed at DDL time by open.ts's ensureSchema, from embed/types.ts's STORE_DIMS) and\n// scans them with array_cosine_similarity, pushing top-k into SQL instead of pulling every row into JS; sqlite's int8+scale BLOB scan is its own representation. Every function here takes `dims` as a parameter rather than importing STORE_DIMS directly, so it stays testable at any width.\n\n// A value import of '@duckdb/node-api' must never sit at module top level, or a sqlite-only tree would resolve this optional dependency\n// just by importing store/index.ts; native.ts's shared duckdbApi() keeps that lazy and reuses whatever open.ts already resolved.\n\n// A bind position whose type is left to auto-inference (safe for plain strings/numbers; only\n// the vector ARRAY positions below need an explicit type -- see writeVectorBatch's comment).\nconst untyped = undefined as unknown as DuckDBType;\n\n// The DDL-fixed array width can exceed a vector's actual length (a hypothetical model sliced under the column's width);\n// zero-padding leaves cosine scores unchanged since the added dimensions contribute nothing to either vector's dot product or norm.\nfunction padded(values: ArrayLike<number>, dims: number): number[] {\n const out = new Array<number>(dims).fill(0);\n for (let d = 0; d < Math.min(values.length, dims); d++) out[d] = values[d];\n return out;\n}\n\nfunction inClause(column: string, count: number): string {\n return `${column} IN (${Array.from({ length: count }, () => '?').join(', ')})`;\n}\n\n// Rows arrive int8-quantized with a per-vector scale (the wire format embed/query.ts produces for every store, see toStore);\n// dequantizing into a plain float array is this store's own representation choice, so both stores are handed the same vectors and diverge only here.\nfunction dequantize(row: VectorWriteRow, dims: number): number[] {\n const q = new Int8Array(row.vector.buffer, row.vector.byteOffset, row.vector.byteLength);\n return padded(\n Array.from(q, (v) => v * row.scale),\n dims\n );\n}\n\n// One runBatch-equivalent call per provider batch: a single multi-row UPDATE (batch.ts's rewriteUpdate), bound directly against the\n// native connection because the vector column needs an explicit ARRAY(DOUBLE, dims) bind type -- auto inference misreads an all-integer first component as INTEGER and silently truncates the rest (verified live: an untyped bind of [0, 0.35, 0.1, -0.2] stores [0,0,0,0]).\nexport async function writeVectorBatch(duckdb: DuckDBConnection, conn: Connection, dims: number, rows: VectorWriteRow[]): Promise<void> {\n if (rows.length === 0) return;\n const { arrayValue, ARRAY, DOUBLE } = await duckdbApi();\n const rewritten = rewriteUpdate('UPDATE embeddings SET vector = ? WHERE \"path\" = ? AND chunk = ?', rows.length);\n if (!rewritten) throw new Error('duckdb: writeVectorBatch SQL shape not recognized by rewriteUpdate');\n\n await withTransaction(conn, async () => {\n const stmt = await duckdb.prepare(rewritten.sql);\n try {\n const values: DuckDBValue[] = [];\n const types: DuckDBType[] = [];\n for (const row of rows) {\n values.push(arrayValue(dequantize(row, dims)), row.path, row.chunk);\n types.push(ARRAY(DOUBLE, dims), untyped, untyped);\n }\n stmt.bind(values, types);\n await stmt.run();\n } finally {\n stmt.destroySync();\n }\n });\n}\n\n// Best chunk per file by cosine, with the earliest chunk winning exact score ties.\n// An empty `allowed` set matches sqlite's scan (every row filtered out) without touching the table.\nexport async function scanCandidates(duckdb: DuckDBConnection, qv: Float32Array, dims: number, fetch: number, allowed?: Set<string>): Promise<VectorCandidate[]> {\n if (allowed && allowed.size === 0) return [];\n const { arrayValue, ARRAY, DOUBLE } = await duckdbApi();\n const allowedList = allowed ? [...allowed] : [];\n const query = padded(qv, dims);\n const queryNorm = Math.sqrt(query.reduce((sum, value) => sum + value * value, 0));\n const score = queryNorm === 0 ? '0.0' : 'CASE WHEN array_inner_product(vector, vector) = 0 THEN 0.0 ELSE array_cosine_similarity(vector, ?) END';\n const sql = `SELECT \"path\", start_line, end_line, score\n FROM (\n SELECT \"path\", start_line, end_line, score,\n row_number() OVER (PARTITION BY \"path\" ORDER BY score DESC, chunk ASC) AS selected\n FROM (\n SELECT \"path\", chunk, start_line, end_line, ${score} AS score\n FROM embeddings\n WHERE vector IS NOT NULL ${allowed ? `AND ${inClause('\"path\"', allowedList.length)}` : ''}\n ) sub\n ) ranked\n WHERE selected = 1\n ORDER BY score DESC, \"path\" ASC\n LIMIT ?`;\n const stmt = await duckdb.prepare(sql);\n try {\n const values: DuckDBValue[] = [...(queryNorm === 0 ? [] : [arrayValue(query)]), ...allowedList, fetch];\n const types: DuckDBType[] = [...(queryNorm === 0 ? [] : [ARRAY(DOUBLE, dims)]), ...allowedList.map(() => untyped), untyped];\n stmt.bind(values, types);\n const reader = await stmt.runAndReadAll();\n const rows = reader.getRowObjectsJS() as Array<{ path: string; start_line: number; end_line: number; score: number }>;\n return rows.map((r) => ({ path: r.path, lines: `L${r.start_line}-${r.end_line}`, similarity: asCosine(r.score) }));\n } finally {\n stmt.destroySync();\n }\n}\n\n// Note-to-note similarity is the max cosine over (target chunk, other chunk) pairs, pushed into one native scan: the sampled target vectors become\n// a small VALUES list, cross-joined against every stored vector and reduced with max()/GROUP BY, so the O(target x stored) cost (12.7s at 201 chunks/note in JS) runs vectorized instead of scalar.\nexport async function scanSimilar(duckdb: DuckDBConnection, conn: Connection, dims: number, path: string, opts: { exclude: Set<string>; allowed?: Set<string>; k: number }): Promise<VectorSimilar[]> {\n if (opts.allowed && opts.allowed.size === 0) return [];\n const targetStmt = await conn.prepare('SELECT vector FROM embeddings WHERE \"path\" = ? AND vector IS NOT NULL ORDER BY chunk');\n const targetRows = (await targetStmt.all(path)) as Array<{ vector: number[] }>;\n if (targetRows.length === 0) return [];\n const targets = sampleEvenly(targetRows).map((row) => row.vector);\n\n const { arrayValue, ARRAY, DOUBLE } = await duckdbApi();\n const excludeList = [...opts.exclude];\n const allowedList = opts.allowed ? [...opts.allowed] : [];\n const sql = `WITH targets(tv) AS (VALUES ${targets.map(() => '(?)').join(', ')}), grouped AS (\n SELECT e.\"path\" AS path, max(CASE WHEN array_inner_product(e.vector, e.vector) = 0 OR array_inner_product(t.tv, t.tv) = 0 THEN 0.0 ELSE array_cosine_similarity(e.vector, t.tv) END) AS score\n FROM embeddings e, targets t\n WHERE e.vector IS NOT NULL AND e.\"path\" != ?\n ${excludeList.length > 0 ? `AND NOT ${inClause('e.\"path\"', excludeList.length)}` : ''}\n ${opts.allowed ? `AND ${inClause('e.\"path\"', allowedList.length)}` : ''}\n GROUP BY e.\"path\"\n )\n SELECT path, score FROM grouped\n ORDER BY score DESC, path ASC\n LIMIT ?`;\n const stmt = await duckdb.prepare(sql);\n try {\n const values: DuckDBValue[] = [...targets.map((t) => arrayValue(padded(t, dims))), path, ...excludeList, ...allowedList, opts.k];\n const types: DuckDBType[] = [...targets.map(() => ARRAY(DOUBLE, dims)), untyped, ...excludeList.map(() => untyped), ...allowedList.map(() => untyped), untyped];\n stmt.bind(values, types);\n const reader = await stmt.runAndReadAll();\n const rows = reader.getRowObjectsJS() as Array<{ path: string; score: number }>;\n return rows.map((r) => ({ path: r.path, similarity: asCosine(r.score) }));\n } finally {\n stmt.destroySync();\n }\n}\n"],"names":["scanCandidates","scanSimilar","writeVectorBatch","untyped","undefined","padded","values","dims","out","Array","fill","d","Math","min","length","inClause","column","count","from","join","dequantize","row","q","Int8Array","vector","buffer","byteOffset","byteLength","v","scale","duckdb","conn","rows","arrayValue","ARRAY","DOUBLE","rewritten","duckdbApi","rewriteUpdate","Error","withTransaction","stmt","types","prepare","sql","push","path","chunk","bind","run","destroySync","qv","fetch","allowed","allowedList","query","queryNorm","score","reader","size","sqrt","reduce","sum","value","map","runAndReadAll","getRowObjectsJS","r","lines","start_line","end_line","similarity","asCosine","opts","targetStmt","targetRows","targets","excludeList","all","sampleEvenly","exclude","t","k"],"mappings":";;;;;;;;;;;QAkEsBA;eAAAA;;QAmCAC;eAAAA;;QA5DAC;eAAAA;;;6BAxCU;yBAEO;uBACT;wBACJ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAE1B,iJAAiJ;AACjJ,+RAA+R;AAE/R,wIAAwI;AACxI,iIAAiI;AAEjI,6FAA6F;AAC7F,6FAA6F;AAC7F,IAAMC,UAAUC;AAEhB,wHAAwH;AACxH,oIAAoI;AACpI,SAASC,OAAOC,MAAyB,EAAEC,IAAY;IACrD,IAAMC,MAAM,IAAIC,MAAcF,MAAMG,IAAI,CAAC;IACzC,IAAK,IAAIC,IAAI,GAAGA,IAAIC,KAAKC,GAAG,CAACP,OAAOQ,MAAM,EAAEP,OAAOI,IAAKH,GAAG,CAACG,EAAE,GAAGL,MAAM,CAACK,EAAE;IAC1E,OAAOH;AACT;AAEA,SAASO,SAASC,MAAc,EAAEC,KAAa;IAC7C,OAAO,AAAC,GAAgBR,OAAdO,QAAO,SAA2D,OAApDP,MAAMS,IAAI,CAAC;QAAEJ,QAAQG;IAAM,GAAG;eAAM;OAAKE,IAAI,CAAC,OAAM;AAC9E;AAEA,6HAA6H;AAC7H,qJAAqJ;AACrJ,SAASC,WAAWC,GAAmB,EAAEd,IAAY;IACnD,IAAMe,IAAI,IAAIC,UAAUF,IAAIG,MAAM,CAACC,MAAM,EAAEJ,IAAIG,MAAM,CAACE,UAAU,EAAEL,IAAIG,MAAM,CAACG,UAAU;IACvF,OAAOtB,OACLI,MAAMS,IAAI,CAACI,GAAG,SAACM;eAAMA,IAAIP,IAAIQ,KAAK;QAClCtB;AAEJ;AAIO,SAAeL,iBAAiB4B,MAAwB,EAAEC,IAAgB,EAAExB,IAAY,EAAEyB,IAAsB;;YAE/E,MAA9BC,YAAYC,OAAOC,QACrBC;;;;oBAFN,IAAIJ,KAAKlB,MAAM,KAAK,GAAG;;;oBACe;;wBAAMuB,IAAAA,mBAAS;;;oBAAf,OAAA,eAA9BJ,aAA8B,KAA9BA,YAAYC,QAAkB,KAAlBA,OAAOC,SAAW,KAAXA;oBACrBC,YAAYE,IAAAA,sBAAa,EAAC,mEAAmEN,KAAKlB,MAAM;oBAC9G,IAAI,CAACsB,WAAW,MAAM,IAAIG,MAAM;oBAEhC;;wBAAMC,IAAAA,8BAAe,EAACT,MAAM;;oCACpBU,MAEEnC,QACAoC,OACD,2BAAA,mBAAA,gBAAA,WAAA,OAAMrB;;;;4CAJA;;gDAAMS,OAAOa,OAAO,CAACP,UAAUQ,GAAG;;;4CAAzCH,OAAO;;;;;;;;;4CAELnC;4CACAoC;4CACD,kCAAA,2BAAA;;gDAAL,IAAK,YAAaV,2BAAb,6BAAA,QAAA,yBAAA,iCAAmB;oDAAbX,MAAN;oDACHf,OAAOuC,IAAI,CAACZ,WAAWb,WAAWC,KAAKd,QAAQc,IAAIyB,IAAI,EAAEzB,IAAI0B,KAAK;oDAClEL,MAAMG,IAAI,CAACX,MAAMC,QAAQ5B,OAAOJ,SAASA;gDAC3C;;gDAHK;gDAAA;;;yDAAA,6BAAA;wDAAA;;;wDAAA;8DAAA;;;;4CAILsC,KAAKO,IAAI,CAAC1C,QAAQoC;4CAClB;;gDAAMD,KAAKQ,GAAG;;;4CAAd;;;;;;4CAEAR,KAAKS,WAAW;;;;;;;;;;4BAEpB;;;;oBAdA;;;;;;IAeF;;AAIO,SAAelD,eAAe8B,MAAwB,EAAEqB,EAAgB,EAAE5C,IAAY,EAAE6C,KAAa,EAAEC,OAAqB;;YAE3F,MAA9BpB,YAAYC,OAAOC,QACrBmB,aACAC,OACAC,WACAC,OACAb,KAaAH,MAEEnC,QACAoC,OAEAgB,QACA1B;;;;oBAzBR,IAAIqB,WAAWA,QAAQM,IAAI,KAAK,GAAG;;;;oBACG;;wBAAMtB,IAAAA,mBAAS;;;oBAAf,OAAA,eAA9BJ,aAA8B,KAA9BA,YAAYC,QAAkB,KAAlBA,OAAOC,SAAW,KAAXA;oBACrBmB,cAAcD,UAAW,qBAAGA;oBAC5BE,QAAQlD,OAAO8C,IAAI5C;oBACnBiD,YAAY5C,KAAKgD,IAAI,CAACL,MAAMM,MAAM,CAAC,SAACC,KAAKC;+BAAUD,MAAMC,QAAQA;uBAAO;oBACxEN,QAAQD,cAAc,IAAI,QAAQ;oBAClCZ,MAAM,AAAC,4QAOoBS,OAFmBI,OAAM,yEAEsC,OAA/DJ,UAAU,AAAC,OAA6C,OAAvCtC,SAAS,UAAUuC,YAAYxC,MAAM,KAAM,IAAG;oBAMnF;;wBAAMgB,OAAOa,OAAO,CAACC;;;oBAA5BH,OAAO;;;;;;;;;oBAELnC,SAAwB,AAAC,qBAAIkD,cAAc;wBAAUvB,WAAWsB;8BAAU,qBAAGD;wBAAaF;;oBAC1FV,QAAsB,AAAC,qBAAIc,cAAc;wBAAUtB,MAAMC,QAAQ5B;8BAAS,qBAAG+C,YAAYU,GAAG,CAAC;+BAAM7D;;wBAAUA;;oBACnHsC,KAAKO,IAAI,CAAC1C,QAAQoC;oBACH;;wBAAMD,KAAKwB,aAAa;;;oBAAjCP,SAAS;oBACT1B,OAAO0B,OAAOQ,eAAe;oBACnC;;wBAAOlC,KAAKgC,GAAG,CAAC,SAACG;mCAAO;gCAAErB,MAAMqB,EAAErB,IAAI;gCAAEsB,OAAO,AAAC,IAAmBD,OAAhBA,EAAEE,UAAU,EAAC,KAAc,OAAXF,EAAEG,QAAQ;gCAAIC,YAAYC,IAAAA,mBAAQ,EAACL,EAAEV,KAAK;4BAAE;;;;oBAE/GhB,KAAKS,WAAW;;;;;;;;;;IAEpB;;AAIO,SAAejD,YAAY6B,MAAwB,EAAEC,IAAgB,EAAExB,IAAY,EAAEuC,IAAY,EAAE2B,IAAgE;;YAElKC,YACAC,YAEAC,SAEgC,MAA9B3C,YAAYC,OAAOC,QACrB0C,aACAvB,aACAV,KAWAH,MAEEnC,QACAoC,OAEAgB,QACA1B;;;;oBA1BR,IAAIyC,KAAKpB,OAAO,IAAIoB,KAAKpB,OAAO,CAACM,IAAI,KAAK,GAAG;;;;oBAC1B;;wBAAM5B,KAAKY,OAAO,CAAC;;;oBAAhC+B,aAAa;oBACC;;wBAAMA,WAAWI,GAAG,CAAChC;;;oBAAnC6B,aAAc;oBACpB,IAAIA,WAAW7D,MAAM,KAAK,GAAG;;;;oBACvB8D,UAAUG,IAAAA,uBAAY,EAACJ,YAAYX,GAAG,CAAC,SAAC3C;+BAAQA,IAAIG,MAAM;;oBAE1B;;wBAAMa,IAAAA,mBAAS;;;oBAAf,OAAA,eAA9BJ,aAA8B,KAA9BA,YAAYC,QAAkB,KAAlBA,OAAOC,SAAW,KAAXA;oBACrB0C,cAAe,qBAAGJ,KAAKO,OAAO;oBAC9B1B,cAAcmB,KAAKpB,OAAO,GAAI,qBAAGoB,KAAKpB,OAAO;oBAC7CT,MAAM,AAAC,+BAIPiC,OAJqCD,QAAQZ,GAAG,CAAC;+BAAM;uBAAO7C,IAAI,CAAC,OAAM,kTAKzEsD,OADAI,YAAY/D,MAAM,GAAG,IAAI,AAAC,WAAmD,OAAzCC,SAAS,YAAY8D,YAAY/D,MAAM,KAAM,IAAG,YACd,OAAtE2D,KAAKpB,OAAO,GAAG,AAAC,OAA+C,OAAzCtC,SAAS,YAAYuC,YAAYxC,MAAM,KAAM,IAAG;oBAM/D;;wBAAMgB,OAAOa,OAAO,CAACC;;;oBAA5BH,OAAO;;;;;;;;;oBAELnC,SAAwB,AAAC,qBAAGsE,QAAQZ,GAAG,CAAC,SAACiB;+BAAMhD,WAAW5B,OAAO4E,GAAG1E;;wBAASuC;uBAAM,qBAAG+B,cAAa,qBAAGvB;wBAAamB,KAAKS,CAAC;;oBACzHxC,QAAsB,AAAC,qBAAGkC,QAAQZ,GAAG,CAAC;+BAAM9B,MAAMC,QAAQ5B;;wBAAQJ;uBAAS,qBAAG0E,YAAYb,GAAG,CAAC;+BAAM7D;yBAAU,qBAAGmD,YAAYU,GAAG,CAAC;+BAAM7D;;wBAAUA;;oBACvJsC,KAAKO,IAAI,CAAC1C,QAAQoC;oBACH;;wBAAMD,KAAKwB,aAAa;;;oBAAjCP,SAAS;oBACT1B,OAAO0B,OAAOQ,eAAe;oBACnC;;wBAAOlC,KAAKgC,GAAG,CAAC,SAACG;mCAAO;gCAAErB,MAAMqB,EAAErB,IAAI;gCAAEyB,YAAYC,IAAAA,mBAAQ,EAACL,EAAEV,KAAK;4BAAE;;;;oBAEtEhB,KAAKS,WAAW;;;;;;;;;;IAEpB"}
|
|
1
|
+
{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/store/duckdb/vectors.ts"],"sourcesContent":["import type { DuckDBConnection, DuckDBType, DuckDBValue } from '@duckdb/node-api';\nimport { withTransaction } from '../transaction.ts';\nimport type { Connection, VectorCandidate, VectorSimilar, VectorWriteRow } from '../types.ts';\nimport { asCosine, sampleEvenly } from '../vectors.ts';\nimport { duckdbApi } from './native.ts';\n\n// This store keeps vectors as native FLOAT[dims] arrays (dims fixed at DDL time by open.ts's ensureSchema, from embed/types.ts's STORE_DIMS) and\n// scans them with array_cosine_similarity, pushing top-k into SQL instead of pulling every row into JS; sqlite's int8+scale BLOB scan is its own representation. Every function here takes `dims` as a parameter rather than importing STORE_DIMS directly, so it stays testable at any width.\n\n// A value import of '@duckdb/node-api' must never sit at module top level, or a sqlite-only tree would resolve this optional dependency\n// just by importing store/index.ts; native.ts's shared duckdbApi() keeps that lazy and reuses whatever open.ts already resolved.\n\n// A bind position whose type is left to auto-inference (safe for plain strings/numbers; only\n// the vector ARRAY positions below need an explicit type -- see writeVectorBatch's comment).\nconst untyped = undefined as unknown as DuckDBType;\nconst VECTOR_WRITE_STAGE = '_sense_vector_write_stage';\n\n// Created once per native connection outside write transactions; its lifetime follows that connection.\nexport async function createVectorWriteStage(duckdb: DuckDBConnection, dims: number): Promise<void> {\n await duckdb.run(`CREATE TEMP TABLE ${VECTOR_WRITE_STAGE} (\"path\" TEXT, chunk INTEGER, vector FLOAT[${dims}])`);\n}\n\n// The DDL-fixed array width can exceed a vector's actual length (a hypothetical model sliced under the column's width);\n// zero-padding leaves cosine scores unchanged since the added dimensions contribute nothing to either vector's dot product or norm.\nfunction padded(values: ArrayLike<number>, dims: number): number[] {\n const out = new Array<number>(dims).fill(0);\n for (let d = 0; d < Math.min(values.length, dims); d++) out[d] = values[d];\n return out;\n}\n\nfunction inClause(column: string, count: number): string {\n return `${column} IN (${Array.from({ length: count }, () => '?').join(', ')})`;\n}\n\n// Rows arrive int8-quantized with a per-vector scale (the wire format embed/query.ts produces for every store, see toStore);\n// dequantizing into a plain float array is this store's own representation choice, so both stores are handed the same vectors and diverge only here.\nfunction dequantize(row: VectorWriteRow, dims: number): number[] {\n const q = new Int8Array(row.vector.buffer, row.vector.byteOffset, row.vector.byteLength);\n return padded(\n Array.from(q, (v) => v * row.scale),\n dims\n );\n}\n\n// The native appender needs an explicit FLOAT array type: inference from an integer first component\n// silently truncates fractional components. The fixed UPDATE keeps each provider batch keyed by path and chunk.\nexport async function writeVectorBatch(duckdb: DuckDBConnection, conn: Connection, dims: number, rows: VectorWriteRow[]): Promise<void> {\n if (rows.length === 0) return;\n const { ARRAY, FLOAT } = await duckdbApi();\n const vectorType = ARRAY(FLOAT, dims);\n\n await withTransaction(conn, async () => {\n await duckdb.run(`DELETE FROM ${VECTOR_WRITE_STAGE}`);\n const appender = await duckdb.createAppender(VECTOR_WRITE_STAGE);\n try {\n for (const row of rows) {\n appender.appendVarchar(row.path);\n appender.appendInteger(row.chunk);\n appender.appendArray(dequantize(row, dims), vectorType);\n appender.endRow();\n }\n appender.flushSync();\n } finally {\n appender.closeSync();\n }\n await duckdb.run(`UPDATE embeddings SET vector = staged.vector FROM ${VECTOR_WRITE_STAGE} AS staged WHERE embeddings.\"path\" = staged.\"path\" AND embeddings.chunk = staged.chunk`);\n });\n}\n\n// Best chunk per file by cosine, with the earliest chunk winning exact score ties.\n// An empty `allowed` set matches sqlite's scan (every row filtered out) without touching the table.\nexport async function scanCandidates(duckdb: DuckDBConnection, qv: Float32Array, dims: number, fetch: number, allowed?: Set<string>): Promise<VectorCandidate[]> {\n if (allowed && allowed.size === 0) return [];\n const { arrayValue, ARRAY, DOUBLE } = await duckdbApi();\n const allowedList = allowed ? [...allowed] : [];\n const query = padded(qv, dims);\n const queryNorm = Math.sqrt(query.reduce((sum, value) => sum + value * value, 0));\n const score = queryNorm === 0 ? '0.0' : 'CASE WHEN array_inner_product(vector, vector) = 0 THEN 0.0 ELSE array_cosine_similarity(vector, ?) END';\n const sql = `SELECT \"path\", start_line, end_line, score\n FROM (\n SELECT \"path\", start_line, end_line, score,\n row_number() OVER (PARTITION BY \"path\" ORDER BY score DESC, chunk ASC) AS selected\n FROM (\n SELECT \"path\", chunk, start_line, end_line, ${score} AS score\n FROM embeddings\n WHERE vector IS NOT NULL ${allowed ? `AND ${inClause('\"path\"', allowedList.length)}` : ''}\n ) sub\n ) ranked\n WHERE selected = 1\n ORDER BY score DESC, \"path\" ASC\n LIMIT ?`;\n const stmt = await duckdb.prepare(sql);\n try {\n const values: DuckDBValue[] = [...(queryNorm === 0 ? [] : [arrayValue(query)]), ...allowedList, fetch];\n const types: DuckDBType[] = [...(queryNorm === 0 ? [] : [ARRAY(DOUBLE, dims)]), ...allowedList.map(() => untyped), untyped];\n stmt.bind(values, types);\n const reader = await stmt.runAndReadAll();\n const rows = reader.getRowObjectsJS() as Array<{ path: string; start_line: number; end_line: number; score: number }>;\n return rows.map((r) => ({ path: r.path, lines: `L${r.start_line}-${r.end_line}`, similarity: asCosine(r.score) }));\n } finally {\n stmt.destroySync();\n }\n}\n\n// Note-to-note similarity is the max cosine over (target chunk, other chunk) pairs, pushed into one native scan: the sampled target vectors become\n// a small VALUES list, cross-joined against every stored vector and reduced with max()/GROUP BY, so the O(target x stored) cost (12.7s at 201 chunks/note in JS) runs vectorized instead of scalar.\nexport async function scanSimilar(duckdb: DuckDBConnection, conn: Connection, dims: number, path: string, opts: { exclude: Set<string>; allowed?: Set<string>; k: number }): Promise<VectorSimilar[]> {\n if (opts.allowed && opts.allowed.size === 0) return [];\n const targetStmt = await conn.prepare('SELECT vector FROM embeddings WHERE \"path\" = ? AND vector IS NOT NULL ORDER BY chunk');\n const targetRows = (await targetStmt.all(path)) as Array<{ vector: number[] }>;\n if (targetRows.length === 0) return [];\n const targets = sampleEvenly(targetRows).map((row) => row.vector);\n\n const { arrayValue, ARRAY, DOUBLE } = await duckdbApi();\n const excludeList = [...opts.exclude];\n const allowedList = opts.allowed ? [...opts.allowed] : [];\n const sql = `WITH targets(tv) AS (VALUES ${targets.map(() => '(?)').join(', ')}), grouped AS (\n SELECT e.\"path\" AS path, max(CASE WHEN array_inner_product(e.vector, e.vector) = 0 OR array_inner_product(t.tv, t.tv) = 0 THEN 0.0 ELSE array_cosine_similarity(e.vector, t.tv) END) AS score\n FROM embeddings e, targets t\n WHERE e.vector IS NOT NULL AND e.\"path\" != ?\n ${excludeList.length > 0 ? `AND NOT ${inClause('e.\"path\"', excludeList.length)}` : ''}\n ${opts.allowed ? `AND ${inClause('e.\"path\"', allowedList.length)}` : ''}\n GROUP BY e.\"path\"\n )\n SELECT path, score FROM grouped\n ORDER BY score DESC, path ASC\n LIMIT ?`;\n const stmt = await duckdb.prepare(sql);\n try {\n const values: DuckDBValue[] = [...targets.map((t) => arrayValue(padded(t, dims))), path, ...excludeList, ...allowedList, opts.k];\n const types: DuckDBType[] = [...targets.map(() => ARRAY(DOUBLE, dims)), untyped, ...excludeList.map(() => untyped), ...allowedList.map(() => untyped), untyped];\n stmt.bind(values, types);\n const reader = await stmt.runAndReadAll();\n const rows = reader.getRowObjectsJS() as Array<{ path: string; score: number }>;\n return rows.map((r) => ({ path: r.path, similarity: asCosine(r.score) }));\n } finally {\n stmt.destroySync();\n }\n}\n"],"names":["createVectorWriteStage","scanCandidates","scanSimilar","writeVectorBatch","untyped","undefined","VECTOR_WRITE_STAGE","duckdb","dims","run","padded","values","out","Array","fill","d","Math","min","length","inClause","column","count","from","join","dequantize","row","q","Int8Array","vector","buffer","byteOffset","byteLength","v","scale","conn","rows","ARRAY","FLOAT","vectorType","duckdbApi","withTransaction","appender","createAppender","appendVarchar","path","appendInteger","chunk","appendArray","endRow","flushSync","closeSync","qv","fetch","allowed","arrayValue","DOUBLE","allowedList","query","queryNorm","score","sql","stmt","types","reader","size","sqrt","reduce","sum","value","prepare","map","bind","runAndReadAll","getRowObjectsJS","r","lines","start_line","end_line","similarity","asCosine","destroySync","opts","targetStmt","targetRows","targets","excludeList","all","sampleEvenly","exclude","t","k"],"mappings":";;;;;;;;;;;QAkBsBA;eAAAA;;QAqDAC;eAAAA;;QAmCAC;eAAAA;;QA5DAC;eAAAA;;;6BA7CU;yBAEO;wBACb;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAE1B,iJAAiJ;AACjJ,+RAA+R;AAE/R,wIAAwI;AACxI,iIAAiI;AAEjI,6FAA6F;AAC7F,6FAA6F;AAC7F,IAAMC,UAAUC;AAChB,IAAMC,qBAAqB;AAGpB,SAAeN,uBAAuBO,MAAwB,EAAEC,IAAY;;;;;oBACjF;;wBAAMD,OAAOE,GAAG,CAAC,AAAC,qBAAoFD,OAAhEF,oBAAmB,+CAAkD,OAALE,MAAK;;;oBAA3G;;;;;;IACF;;AAEA,wHAAwH;AACxH,oIAAoI;AACpI,SAASE,OAAOC,MAAyB,EAAEH,IAAY;IACrD,IAAMI,MAAM,IAAIC,MAAcL,MAAMM,IAAI,CAAC;IACzC,IAAK,IAAIC,IAAI,GAAGA,IAAIC,KAAKC,GAAG,CAACN,OAAOO,MAAM,EAAEV,OAAOO,IAAKH,GAAG,CAACG,EAAE,GAAGJ,MAAM,CAACI,EAAE;IAC1E,OAAOH;AACT;AAEA,SAASO,SAASC,MAAc,EAAEC,KAAa;IAC7C,OAAO,AAAC,GAAgBR,OAAdO,QAAO,SAA2D,OAApDP,MAAMS,IAAI,CAAC;QAAEJ,QAAQG;IAAM,GAAG;eAAM;OAAKE,IAAI,CAAC,OAAM;AAC9E;AAEA,6HAA6H;AAC7H,qJAAqJ;AACrJ,SAASC,WAAWC,GAAmB,EAAEjB,IAAY;IACnD,IAAMkB,IAAI,IAAIC,UAAUF,IAAIG,MAAM,CAACC,MAAM,EAAEJ,IAAIG,MAAM,CAACE,UAAU,EAAEL,IAAIG,MAAM,CAACG,UAAU;IACvF,OAAOrB,OACLG,MAAMS,IAAI,CAACI,GAAG,SAACM;eAAMA,IAAIP,IAAIQ,KAAK;QAClCzB;AAEJ;AAIO,SAAeL,iBAAiBI,MAAwB,EAAE2B,IAAgB,EAAE1B,IAAY,EAAE2B,IAAsB;;YAE5F,MAAjBC,OAAOC,OACTC;;;;oBAFN,IAAIH,KAAKjB,MAAM,KAAK,GAAG;;;oBACE;;wBAAMqB,IAAAA,mBAAS;;;oBAAf,OAAA,eAAjBH,QAAiB,KAAjBA,OAAOC,QAAU,KAAVA;oBACTC,aAAaF,MAAMC,OAAO7B;oBAEhC;;wBAAMgC,IAAAA,8BAAe,EAACN,MAAM;;oCAEpBO,UAEC,2BAAA,mBAAA,gBAAA,WAAA,OAAMhB;;;;4CAHb;;gDAAMlB,OAAOE,GAAG,CAAC,AAAC,eAAiC,OAAnBH;;;4CAAhC;4CACiB;;gDAAMC,OAAOmC,cAAc,CAACpC;;;4CAAvCmC,WAAW;4CACjB,IAAI;gDACG,kCAAA,2BAAA;;oDAAL,IAAK,YAAaN,2BAAb,6BAAA,QAAA,yBAAA,iCAAmB;wDAAbV,MAAN;wDACHgB,SAASE,aAAa,CAAClB,IAAImB,IAAI;wDAC/BH,SAASI,aAAa,CAACpB,IAAIqB,KAAK;wDAChCL,SAASM,WAAW,CAACvB,WAAWC,KAAKjB,OAAO8B;wDAC5CG,SAASO,MAAM;oDACjB;;oDALK;oDAAA;;;6DAAA,6BAAA;4DAAA;;;4DAAA;kEAAA;;;;gDAMLP,SAASQ,SAAS;4CACpB,SAAU;gDACRR,SAASS,SAAS;4CACpB;4CACA;;gDAAM3C,OAAOE,GAAG,CAAC,AAAC,qDAAuE,OAAnBH,oBAAmB;;;4CAAzF;;;;;;4BACF;;;;oBAfA;;;;;;IAgBF;;AAIO,SAAeL,eAAeM,MAAwB,EAAE4C,EAAgB,EAAE3C,IAAY,EAAE4C,KAAa,EAAEC,OAAqB;;YAE3F,MAA9BC,YAAYlB,OAAOmB,QACrBC,aACAC,OACAC,WACAC,OACAC,KAaAC,MAEElD,QACAmD,OAEAC,QACA5B;;;;oBAzBR,IAAIkB,WAAWA,QAAQW,IAAI,KAAK,GAAG;;;;oBACG;;wBAAMzB,IAAAA,mBAAS;;;oBAAf,OAAA,eAA9Be,aAA8B,KAA9BA,YAAYlB,QAAkB,KAAlBA,OAAOmB,SAAW,KAAXA;oBACrBC,cAAcH,UAAW,qBAAGA;oBAC5BI,QAAQ/C,OAAOyC,IAAI3C;oBACnBkD,YAAY1C,KAAKiD,IAAI,CAACR,MAAMS,MAAM,CAAC,SAACC,KAAKC;+BAAUD,MAAMC,QAAQA;uBAAO;oBACxET,QAAQD,cAAc,IAAI,QAAQ;oBAClCE,MAAM,AAAC,4QAOoBP,OAFmBM,OAAM,yEAEsC,OAA/DN,UAAU,AAAC,OAA6C,OAAvClC,SAAS,UAAUqC,YAAYtC,MAAM,KAAM,IAAG;oBAMnF;;wBAAMX,OAAO8D,OAAO,CAACT;;;oBAA5BC,OAAO;;;;;;;;;oBAELlD,SAAwB,AAAC,qBAAI+C,cAAc;wBAAUJ,WAAWG;8BAAU,qBAAGD;wBAAaJ;;oBAC1FU,QAAsB,AAAC,qBAAIJ,cAAc;wBAAUtB,MAAMmB,QAAQ/C;8BAAS,qBAAGgD,YAAYc,GAAG,CAAC;+BAAMlE;;wBAAUA;;oBACnHyD,KAAKU,IAAI,CAAC5D,QAAQmD;oBACH;;wBAAMD,KAAKW,aAAa;;;oBAAjCT,SAAS;oBACT5B,OAAO4B,OAAOU,eAAe;oBACnC;;wBAAOtC,KAAKmC,GAAG,CAAC,SAACI;mCAAO;gCAAE9B,MAAM8B,EAAE9B,IAAI;gCAAE+B,OAAO,AAAC,IAAmBD,OAAhBA,EAAEE,UAAU,EAAC,KAAc,OAAXF,EAAEG,QAAQ;gCAAIC,YAAYC,IAAAA,mBAAQ,EAACL,EAAEf,KAAK;4BAAE;;;;oBAE/GE,KAAKmB,WAAW;;;;;;;;;;IAEpB;;AAIO,SAAe9E,YAAYK,MAAwB,EAAE2B,IAAgB,EAAE1B,IAAY,EAAEoC,IAAY,EAAEqC,IAAgE;;YAElKC,YACAC,YAEAC,SAEgC,MAA9B9B,YAAYlB,OAAOmB,QACrB8B,aACA7B,aACAI,KAWAC,MAEElD,QACAmD,OAEAC,QACA5B;;;;oBA1BR,IAAI8C,KAAK5B,OAAO,IAAI4B,KAAK5B,OAAO,CAACW,IAAI,KAAK,GAAG;;;;oBAC1B;;wBAAM9B,KAAKmC,OAAO,CAAC;;;oBAAhCa,aAAa;oBACC;;wBAAMA,WAAWI,GAAG,CAAC1C;;;oBAAnCuC,aAAc;oBACpB,IAAIA,WAAWjE,MAAM,KAAK,GAAG;;;;oBACvBkE,UAAUG,IAAAA,uBAAY,EAACJ,YAAYb,GAAG,CAAC,SAAC7C;+BAAQA,IAAIG,MAAM;;oBAE1B;;wBAAMW,IAAAA,mBAAS;;;oBAAf,OAAA,eAA9Be,aAA8B,KAA9BA,YAAYlB,QAAkB,KAAlBA,OAAOmB,SAAW,KAAXA;oBACrB8B,cAAe,qBAAGJ,KAAKO,OAAO;oBAC9BhC,cAAcyB,KAAK5B,OAAO,GAAI,qBAAG4B,KAAK5B,OAAO;oBAC7CO,MAAM,AAAC,+BAIPyB,OAJqCD,QAAQd,GAAG,CAAC;+BAAM;uBAAO/C,IAAI,CAAC,OAAM,kTAKzE0D,OADAI,YAAYnE,MAAM,GAAG,IAAI,AAAC,WAAmD,OAAzCC,SAAS,YAAYkE,YAAYnE,MAAM,KAAM,IAAG,YACd,OAAtE+D,KAAK5B,OAAO,GAAG,AAAC,OAA+C,OAAzClC,SAAS,YAAYqC,YAAYtC,MAAM,KAAM,IAAG;oBAM/D;;wBAAMX,OAAO8D,OAAO,CAACT;;;oBAA5BC,OAAO;;;;;;;;;oBAELlD,SAAwB,AAAC,qBAAGyE,QAAQd,GAAG,CAAC,SAACmB;+BAAMnC,WAAW5C,OAAO+E,GAAGjF;;wBAASoC;uBAAM,qBAAGyC,cAAa,qBAAG7B;wBAAayB,KAAKS,CAAC;;oBACzH5B,QAAsB,AAAC,qBAAGsB,QAAQd,GAAG,CAAC;+BAAMlC,MAAMmB,QAAQ/C;;wBAAQJ;uBAAS,qBAAGiF,YAAYf,GAAG,CAAC;+BAAMlE;yBAAU,qBAAGoD,YAAYc,GAAG,CAAC;+BAAMlE;;wBAAUA;;oBACvJyD,KAAKU,IAAI,CAAC5D,QAAQmD;oBACH;;wBAAMD,KAAKW,aAAa;;;oBAAjCT,SAAS;oBACT5B,OAAO4B,OAAOU,eAAe;oBACnC;;wBAAOtC,KAAKmC,GAAG,CAAC,SAACI;mCAAO;gCAAE9B,MAAM8B,EAAE9B,IAAI;gCAAEkC,YAAYC,IAAAA,mBAAQ,EAACL,EAAEf,KAAK;4BAAE;;;;oBAEtEE,KAAKmB,WAAW;;;;;;;;;;IAEpB"}
|
package/dist/cjs/watch.js
CHANGED
|
@@ -9,10 +9,10 @@ Object.defineProperty(exports, "runWatch", {
|
|
|
9
9
|
}
|
|
10
10
|
});
|
|
11
11
|
var _nodefs = require("node:fs");
|
|
12
|
-
var _indexts = require("./config/index.js");
|
|
13
12
|
var _errorsts = require("./errors.js");
|
|
14
13
|
var _guardedtickts = require("./lib/guarded-tick.js");
|
|
15
|
-
var
|
|
14
|
+
var _watchnotificationts = require("./lib/watch-notification.js");
|
|
15
|
+
var _indexts = require("./store/index.js");
|
|
16
16
|
var _watchclaimts = require("./watch-claim.js");
|
|
17
17
|
function asyncGeneratorStep(gen, resolve, reject, _next, _throw, key, arg) {
|
|
18
18
|
try {
|
|
@@ -188,7 +188,7 @@ function runWatch(_0) {
|
|
|
188
188
|
]);
|
|
189
189
|
return [
|
|
190
190
|
4,
|
|
191
|
-
(0,
|
|
191
|
+
(0, _indexts.openStore)(cfg)
|
|
192
192
|
];
|
|
193
193
|
case 3:
|
|
194
194
|
_ref = _state.sent(), initialStore = _ref.store, dbPath = _ref.dbPath, initialWarnings = _ref.warnings, initialParsed = _ref.parsed;
|
|
@@ -207,7 +207,7 @@ function runWatch(_0) {
|
|
|
207
207
|
];
|
|
208
208
|
return [
|
|
209
209
|
4,
|
|
210
|
-
(0,
|
|
210
|
+
(0, _indexts.docCount)(initialStore)
|
|
211
211
|
];
|
|
212
212
|
case 5:
|
|
213
213
|
_tmp = _state.sent();
|
|
@@ -256,7 +256,7 @@ function runWatch(_0) {
|
|
|
256
256
|
]);
|
|
257
257
|
return [
|
|
258
258
|
4,
|
|
259
|
-
(0,
|
|
259
|
+
(0, _indexts.openStore)(cfg)
|
|
260
260
|
];
|
|
261
261
|
case 1:
|
|
262
262
|
_ref = _state.sent(), store = _ref.store, parsed = _ref.parsed, warnings = _ref.warnings;
|
|
@@ -278,7 +278,7 @@ function runWatch(_0) {
|
|
|
278
278
|
};
|
|
279
279
|
return [
|
|
280
280
|
4,
|
|
281
|
-
(0,
|
|
281
|
+
(0, _indexts.docCount)(store)
|
|
282
282
|
];
|
|
283
283
|
case 3:
|
|
284
284
|
onEvent.apply(void 0, [
|
|
@@ -354,7 +354,7 @@ function runWatch(_0) {
|
|
|
354
354
|
watcher = (0, _nodefs.watch)(rootDir, {
|
|
355
355
|
recursive: true
|
|
356
356
|
}, function(_event, filename) {
|
|
357
|
-
if (
|
|
357
|
+
if (!(0, _watchnotificationts.shouldReconcileWatchNotification)(filename)) return;
|
|
358
358
|
scheduleReconcile();
|
|
359
359
|
});
|
|
360
360
|
reconcileTimer = setInterval((0, _guardedtickts.guardedTick)(function() {
|
|
@@ -410,7 +410,7 @@ function runWatch(_0) {
|
|
|
410
410
|
];
|
|
411
411
|
return [
|
|
412
412
|
4,
|
|
413
|
-
(0,
|
|
413
|
+
(0, _indexts.openStore)(cfg)
|
|
414
414
|
];
|
|
415
415
|
case 3:
|
|
416
416
|
_ref = _state.sent(), finalStore = _ref.store;
|
package/dist/cjs/watch.js.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/watch.ts"],"sourcesContent":["import { watch as fsWatch } from 'node:fs';\nimport type { ResolvedConfig } from './config/index.ts';\nimport { STATE_DIR } from './config/index.ts';\nimport { SenseError } from './errors.ts';\nimport { guardedTick } from './lib/guarded-tick.ts';\nimport { docCount, openStore } from './store/index.ts';\nimport { startWatchClaim, WATCH_HEARTBEAT_INTERVAL_MS } from './watch-claim.ts';\n\n// Watch is a cache pre-warmer, not a correctness mechanism: open() always reconciles anyway, so any fs event just triggers a debounced full reconcile.\nconst DEBOUNCE_MS = 200;\n\nexport type WatchEvent = { type: 'started'; rootDir: string; dbPath: string } | { type: 'reconciled'; parsed: number; total: number; warnings: string[] } | { type: 'reconcile-error'; message: string };\n\nexport interface WatchOptions {\n force?: boolean;\n onEvent?: (event: WatchEvent) => void;\n // Aborting runs the same shutdown path as SIGINT/SIGTERM.\n signal?: AbortSignal;\n debounceMs?: number;\n heartbeatIntervalMs?: number;\n}\n\n// Runs in the foreground until SIGINT/SIGTERM/signal abort. DuckDB and Turso hand the cache file to\n// one process at a time, so every reconcile cycle opens and closes instead of holding an idle lock.\nexport async function runWatch(cfg: ResolvedConfig, opts: WatchOptions = {}): Promise<void> {\n const onEvent = opts.onEvent ?? (() => {});\n const debounceMs = opts.debounceMs ?? DEBOUNCE_MS;\n const reconcileIntervalMs = opts.heartbeatIntervalMs ?? WATCH_HEARTBEAT_INTERVAL_MS;\n if (!Number.isFinite(reconcileIntervalMs) || reconcileIntervalMs <= 0) throw new SenseError('CONFIG_INVALID', 'watch heartbeatIntervalMs must be a finite positive number');\n // Reconciliation may be parked for a caller, but live ownership must renew before fixed expiry.\n const claimIntervalMs = Math.min(reconcileIntervalMs, WATCH_HEARTBEAT_INTERVAL_MS);\n const rootDir = cfg.rootDir ?? cfg.baseDir;\n const configDir = cfg.configDir ?? rootDir;\n let claimFailure: Error | null = null;\n let requestShutdown: ((reason: Error) => void) | undefined;\n const claim = await startWatchClaim(configDir, opts.force ?? false, claimIntervalMs, (err) => {\n claimFailure ??= err;\n requestShutdown?.(err);\n });\n let claimCloseStarted = false;\n const closeClaim = (): Promise<void> => {\n claimCloseStarted = true;\n return claim.close();\n };\n try {\n const { store: initialStore, dbPath, warnings: initialWarnings, parsed: initialParsed } = await openStore(cfg);\n let initialTotal = 0;\n try {\n // total is read before close: an event fires only once nothing is held open again.\n initialTotal = initialWarnings.length > 0 || initialParsed > 0 ? await docCount(initialStore) : 0;\n } finally {\n await initialStore.close();\n }\n if (claimFailure) throw claimFailure;\n\n let stopping = false;\n // At most one open+reconcile+close cycle runs at a time; a trigger that arrives mid-cycle is\n // coalesced into a single rerun instead of opening a second, overlapping connection.\n let running: Promise<void> | null = null;\n let queuedAlways: boolean | null = null;\n\n const tick = async (alwaysEmit: boolean): Promise<void> => {\n try {\n const { store, parsed, warnings } = await openStore(cfg);\n try {\n if (alwaysEmit || parsed > 0 || warnings.length > 0) {\n onEvent({ type: 'reconciled', parsed, total: await docCount(store), warnings });\n }\n } finally {\n await store.close();\n }\n } catch (err) {\n onEvent({ type: 'reconcile-error', message: (err as Error).message });\n }\n };\n\n const startCycle = (alwaysEmit: boolean) => {\n running = tick(alwaysEmit).then(() => {\n running = null;\n if (queuedAlways !== null && !stopping) {\n const again = queuedAlways;\n queuedAlways = null;\n startCycle(again);\n }\n });\n };\n\n const requestCycle = (alwaysEmit: boolean) => {\n if (stopping) return;\n if (running) {\n queuedAlways = queuedAlways || alwaysEmit;\n return;\n }\n startCycle(alwaysEmit);\n };\n\n let debounceTimer: NodeJS.Timeout | null = null;\n const scheduleReconcile = () => {\n if (debounceTimer) clearTimeout(debounceTimer);\n debounceTimer = setTimeout(() => {\n debounceTimer = null;\n requestCycle(true);\n }, debounceMs);\n };\n\n // Ignore our own state files, or cache and claim writes would retrigger reconciliation forever.\n // An unresolvable filename reconciles because parsing nothing costs less than missing a real edit.\n const watcher = fsWatch(rootDir, { recursive: true }, (_event, filename) => {\n if (typeof filename === 'string' && filename.startsWith(STATE_DIR)) return;\n scheduleReconcile();\n });\n const reconcileTimer = setInterval(\n guardedTick(\n async () => requestCycle(false),\n () => stopping\n ),\n reconcileIntervalMs\n );\n\n return await new Promise<void>((resolveShutdown, rejectShutdown) => {\n let shutdownReason: Error | null = null;\n const shutdown = (reason?: Error) => {\n shutdownReason ??= reason ?? null;\n if (stopping) return;\n stopping = true;\n const shutdownTask = (async () => {\n process.off('SIGINT', stopNormally);\n process.off('SIGTERM', stopNormally);\n opts.signal?.removeEventListener('abort', stopNormally);\n clearInterval(reconcileTimer);\n if (debounceTimer) clearTimeout(debounceTimer);\n watcher.close();\n await running;\n let finalError: unknown;\n try {\n // A normal stop performs one final freshness pass. Ownership loss drains existing work\n // but starts nothing new because the replacement watcher now owns that responsibility.\n if (!shutdownReason) {\n const { store: finalStore } = await openStore(cfg);\n await finalStore.close();\n }\n } catch (err) {\n finalError = err;\n }\n try {\n await closeClaim();\n } catch (err) {\n if (finalError) throw new AggregateError([finalError, err], 'watch shutdown and claim cleanup both failed');\n throw err;\n }\n if (finalError) throw finalError;\n if (shutdownReason) throw shutdownReason;\n })();\n shutdownTask.then(resolveShutdown, rejectShutdown);\n };\n const stopNormally = () => shutdown();\n requestShutdown = shutdown;\n process.once('SIGINT', stopNormally);\n process.once('SIGTERM', stopNormally);\n if (claimFailure) shutdown(claimFailure);\n else if (opts.signal?.aborted) shutdown();\n else {\n opts.signal?.addEventListener('abort', stopNormally, { once: true });\n try {\n onEvent({ type: 'started', rootDir, dbPath });\n if (!stopping && (initialWarnings.length > 0 || initialParsed > 0)) {\n onEvent({ type: 'reconciled', parsed: initialParsed, total: initialTotal, warnings: initialWarnings });\n }\n } catch (err) {\n shutdown(err instanceof Error ? err : new Error(String(err)));\n }\n }\n });\n } catch (err) {\n if (claimCloseStarted) throw err;\n try {\n await closeClaim();\n } catch (closeError) {\n throw new AggregateError([err, closeError], 'watch setup and claim cleanup both failed');\n }\n throw err;\n } finally {\n requestShutdown = undefined;\n }\n}\n"],"names":["runWatch","DEBOUNCE_MS","cfg","opts","onEvent","debounceMs","reconcileIntervalMs","claimIntervalMs","rootDir","configDir","claimFailure","requestShutdown","claim","claimCloseStarted","closeClaim","initialStore","dbPath","initialWarnings","initialParsed","initialTotal","stopping","running","queuedAlways","tick","startCycle","requestCycle","debounceTimer","scheduleReconcile","watcher","reconcileTimer","err","closeError","heartbeatIntervalMs","WATCH_HEARTBEAT_INTERVAL_MS","Number","isFinite","SenseError","Math","min","baseDir","startWatchClaim","force","close","openStore","store","warnings","parsed","length","docCount","alwaysEmit","type","total","message","then","again","clearTimeout","setTimeout","fsWatch","recursive","_event","filename","startsWith","STATE_DIR","setInterval","guardedTick","Promise","resolveShutdown","rejectShutdown","shutdownReason","shutdown","reason","shutdownTask","finalError","finalStore","process","off","stopNormally","signal","removeEventListener","clearInterval","AggregateError","once","aborted","addEventListener","Error","String","undefined"],"mappings":";;;;+BAwBsBA;;;eAAAA;;;sBAxBW;uBAEP;wBACC;6BACC;wBACQ;4BACyB;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAE7D,uJAAuJ;AACvJ,IAAMC,cAAc;AAeb,SAAeD;wCAASE,GAAmB;YAAEC,MAClCA,eACGA,kBACSA,2BAIZD,cACEA,gBAG6BC,aAVzCC,SACAC,YACAC,qBAGAC,iBACAC,SACAC,WACFC,cACAC,iBACEC,OAIFC,mBACEC,YAKsF,MAA3EC,cAAcC,QAAkBC,iBAAyBC,eACpEC,oBASAC,UAGAC,SACAC,cAEEC,MAeAC,YAWAC,cASFC,eACEC,mBAUAC,SAIAC,gBA8DCC,KAIEC;;;;;oBAzJuC5B,OAAAA,oEAAqB,CAAC;oBAClEC,WAAUD,gBAAAA,KAAKC,OAAO,cAAZD,2BAAAA,gBAAiB,YAAO;oBAClCE,cAAaF,mBAAAA,KAAKE,UAAU,cAAfF,8BAAAA,mBAAmBF;oBAChCK,uBAAsBH,4BAAAA,KAAK6B,mBAAmB,cAAxB7B,uCAAAA,4BAA4B8B,yCAA2B;oBACnF,IAAI,CAACC,OAAOC,QAAQ,CAAC7B,wBAAwBA,uBAAuB,GAAG,MAAM,IAAI8B,oBAAU,CAAC,kBAAkB;oBAC9G,gGAAgG;oBAC1F7B,kBAAkB8B,KAAKC,GAAG,CAAChC,qBAAqB2B,yCAA2B;oBAC3EzB,WAAUN,eAAAA,IAAIM,OAAO,cAAXN,0BAAAA,eAAeA,IAAIqC,OAAO;oBACpC9B,aAAYP,iBAAAA,IAAIO,SAAS,cAAbP,4BAAAA,iBAAiBM;oBAC/BE,eAA6B;oBAEnB;;wBAAM8B,IAAAA,6BAAe,EAAC/B,YAAWN,cAAAA,KAAKsC,KAAK,cAAVtC,yBAAAA,cAAc,OAAOI,iBAAiB,SAACuB;4BACpFpB,yBAAAA,0BAAAA,eAAAA,eAAiBoB;4BACjBnB,4BAAAA,sCAAAA,gBAAkBmB;wBACpB;;;oBAHMlB,QAAQ;oBAIVC,oBAAoB;oBAClBC,aAAa;wBACjBD,oBAAoB;wBACpB,OAAOD,MAAM8B,KAAK;oBACpB;;;;;;;;;oBAE4F;;wBAAMC,IAAAA,mBAAS,EAACzC;;;oBAAhB,OAAA,eAA3Ea,eAA2E,KAAlF6B,OAAqB5B,SAA6D,KAA7DA,QAAkBC,kBAA2C,KAArD4B,UAAmC3B,gBAAkB,KAA1B4B;oBAC5D3B,eAAe;;;;;;;;;yBAGFF,CAAAA,gBAAgB8B,MAAM,GAAG,KAAK7B,gBAAgB,CAAA,GAA9CD;;;;oBAAkD;;wBAAM+B,IAAAA,kBAAQ,EAACjC;;;2BAAf;;;;;;2BAA+B;;;oBADhG,mFAAmF;oBACnFI;;;;;;oBAEA;;wBAAMJ,aAAa2B,KAAK;;;oBAAxB;;;;;oBAEF,IAAIhC,cAAc,MAAMA;oBAEpBU,WAAW;oBACf,6FAA6F;oBAC7F,qFAAqF;oBACjFC,UAAgC;oBAChCC,eAA+B;oBAE7BC,OAAO,cAAO0B;;gCAEoB,MAA5BL,OAAOE,QAAQD,gBAQhBf;;;;;;;;;;wCAR6B;;4CAAMa,IAAAA,mBAAS,EAACzC;;;wCAAhB,OAAA,eAA5B0C,QAA4B,KAA5BA,OAAOE,SAAqB,KAArBA,QAAQD,WAAa,KAAbA;;;;;;;;;6CAEjBI,CAAAA,cAAcH,SAAS,KAAKD,SAASE,MAAM,GAAG,CAAA,GAA9CE;;;;;4CACQC,MAAM;4CAAcJ,QAAAA;;wCAAe;;4CAAME,IAAAA,kBAAQ,EAACJ;;;wCAA5DxC;kDAAsC+C,QAAO,oBAAuBN,WAAAA;;;;;;;;;wCAGtE;;4CAAMD,MAAMF,KAAK;;;wCAAjB;;;;;;;;;;wCAEKZ;wCACP1B,QAAQ;4CAAE8C,MAAM;4CAAmBE,SAAS,AAACtB,IAAcsB,OAAO;wCAAC;;;;;;;;;;;wBAEvE;;oBAEM5B,aAAa,qBAACyB;wBAClB5B,UAAUE,KAAK0B,YAAYI,IAAI,CAAC;4BAC9BhC,UAAU;4BACV,IAAIC,iBAAiB,QAAQ,CAACF,UAAU;gCACtC,IAAMkC,QAAQhC;gCACdA,eAAe;gCACfE,WAAW8B;4BACb;wBACF;oBACF;oBAEM7B,eAAe,sBAACwB;wBACpB,IAAI7B,UAAU;wBACd,IAAIC,SAAS;4BACXC,eAAeA,gBAAgB2B;4BAC/B;wBACF;wBACAzB,WAAWyB;oBACb;oBAEIvB,gBAAuC;oBACrCC,oBAAoB;wBACxB,IAAID,eAAe6B,aAAa7B;wBAChCA,gBAAgB8B,WAAW;4BACzB9B,gBAAgB;4BAChBD,aAAa;wBACf,GAAGpB;oBACL;oBAEA,gGAAgG;oBAChG,mGAAmG;oBAC7FuB,UAAU6B,IAAAA,aAAO,EAACjD,SAAS;wBAAEkD,WAAW;oBAAK,GAAG,SAACC,QAAQC;wBAC7D,IAAI,OAAOA,aAAa,YAAYA,SAASC,UAAU,CAACC,kBAAS,GAAG;wBACpEnC;oBACF;oBACME,iBAAiBkC,YACrBC,IAAAA,0BAAW,EACT;;;;;oCAAYvC,aAAa;;;;uBACzB;+BAAML;wBAERd;oBAGK;;wBAAM,IAAI2D,QAAc,SAACC,iBAAiBC;gCAyCtChE;4BAxCT,IAAIiE,iBAA+B;4BACnC,IAAMC,WAAW,kBAACC;gCAChBF,2BAAAA,4BAAAA,iBAAAA,iBAAmBE,mBAAAA,oBAAAA,SAAU;gCAC7B,IAAIlD,UAAU;gCACdA,WAAW;gCACX,IAAMmD,eAAe,AAAC;;4CAGpBpE,cAKIqE,YAK8B,MAAfC,YAGV3C,KAKAA;;;;oDApBT4C,QAAQC,GAAG,CAAC,UAAUC;oDACtBF,QAAQC,GAAG,CAAC,WAAWC;qDACvBzE,eAAAA,KAAK0E,MAAM,cAAX1E,mCAAAA,aAAa2E,mBAAmB,CAAC,SAASF;oDAC1CG,cAAclD;oDACd,IAAIH,eAAe6B,aAAa7B;oDAChCE,QAAQc,KAAK;oDACb;;wDAAMrB;;;oDAAN;;;;;;;;;yDAKM,CAAC+C,gBAAD;;;;oDAC4B;;wDAAMzB,IAAAA,mBAAS,EAACzC;;;oDAAhB,OAAA,eAAfuE,aAAe,KAAtB7B;oDACR;;wDAAM6B,WAAW/B,KAAK;;;oDAAtB;;;;;;;;oDAEKZ;oDACP0C,aAAa1C;;;;;;;;;;;;oDAGb;;wDAAMhB;;;oDAAN;;;;;;oDACOgB;oDACP,IAAI0C,YAAY,MAAM,IAAIQ;wDAAgBR;wDAAY1C;uDAAM;oDAC5D,MAAMA;;oDAER,IAAI0C,YAAY,MAAMA;oDACtB,IAAIJ,gBAAgB,MAAMA;;;;;;oCAC5B;;gCACAG,aAAalB,IAAI,CAACa,iBAAiBC;4BACrC;4BACA,IAAMS,eAAe;uCAAMP;;4BAC3B1D,kBAAkB0D;4BAClBK,QAAQO,IAAI,CAAC,UAAUL;4BACvBF,QAAQO,IAAI,CAAC,WAAWL;4BACxB,IAAIlE,cAAc2D,SAAS3D;iCACtB,KAAIP,eAAAA,KAAK0E,MAAM,cAAX1E,mCAAAA,aAAa+E,OAAO,EAAEb;iCAC1B;oCACHlE;iCAAAA,gBAAAA,KAAK0E,MAAM,cAAX1E,oCAAAA,cAAagF,gBAAgB,CAAC,SAASP,cAAc;oCAAEK,MAAM;gCAAK;gCAClE,IAAI;oCACF7E,QAAQ;wCAAE8C,MAAM;wCAAW1C,SAAAA;wCAASQ,QAAAA;oCAAO;oCAC3C,IAAI,CAACI,YAAaH,CAAAA,gBAAgB8B,MAAM,GAAG,KAAK7B,gBAAgB,CAAA,GAAI;wCAClEd,QAAQ;4CAAE8C,MAAM;4CAAcJ,QAAQ5B;4CAAeiC,OAAOhC;4CAAc0B,UAAU5B;wCAAgB;oCACtG;gCACF,EAAE,OAAOa,KAAK;oCACZuC,SAASvC,AAAG,YAAHA,KAAesD,SAAQtD,MAAM,IAAIsD,MAAMC,OAAOvD;gCACzD;4BACF;wBACF;;;oBArDA;;wBAAO;;;oBAsDAA;oBACP,IAAIjB,mBAAmB,MAAMiB;;;;;;;;;oBAE3B;;wBAAMhB;;;oBAAN;;;;;;oBACOiB;oBACP,MAAM,IAAIiD;wBAAgBlD;wBAAKC;uBAAa;;oBAE9C,MAAMD;;oBAENnB,kBAAkB2E;;;;;;;;;;IAEtB"}
|
|
1
|
+
{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/watch.ts"],"sourcesContent":["import { watch as fsWatch } from 'node:fs';\nimport type { ResolvedConfig } from './config/index.ts';\nimport { SenseError } from './errors.ts';\nimport { guardedTick } from './lib/guarded-tick.ts';\nimport { shouldReconcileWatchNotification } from './lib/watch-notification.ts';\nimport { docCount, openStore } from './store/index.ts';\nimport { startWatchClaim, WATCH_HEARTBEAT_INTERVAL_MS } from './watch-claim.ts';\n\n// Watch is a cache pre-warmer, not a correctness mechanism: open() always reconciles anyway, so any fs event just triggers a debounced full reconcile.\nconst DEBOUNCE_MS = 200;\n\nexport type WatchEvent = { type: 'started'; rootDir: string; dbPath: string } | { type: 'reconciled'; parsed: number; total: number; warnings: string[] } | { type: 'reconcile-error'; message: string };\n\nexport interface WatchOptions {\n force?: boolean;\n onEvent?: (event: WatchEvent) => void;\n // Aborting runs the same shutdown path as SIGINT/SIGTERM.\n signal?: AbortSignal;\n debounceMs?: number;\n heartbeatIntervalMs?: number;\n}\n\n// Runs in the foreground until SIGINT/SIGTERM/signal abort. DuckDB and Turso hand the cache file to\n// one process at a time, so every reconcile cycle opens and closes instead of holding an idle lock.\nexport async function runWatch(cfg: ResolvedConfig, opts: WatchOptions = {}): Promise<void> {\n const onEvent = opts.onEvent ?? (() => {});\n const debounceMs = opts.debounceMs ?? DEBOUNCE_MS;\n const reconcileIntervalMs = opts.heartbeatIntervalMs ?? WATCH_HEARTBEAT_INTERVAL_MS;\n if (!Number.isFinite(reconcileIntervalMs) || reconcileIntervalMs <= 0) throw new SenseError('CONFIG_INVALID', 'watch heartbeatIntervalMs must be a finite positive number');\n // Reconciliation may be parked for a caller, but live ownership must renew before fixed expiry.\n const claimIntervalMs = Math.min(reconcileIntervalMs, WATCH_HEARTBEAT_INTERVAL_MS);\n const rootDir = cfg.rootDir ?? cfg.baseDir;\n const configDir = cfg.configDir ?? rootDir;\n let claimFailure: Error | null = null;\n let requestShutdown: ((reason: Error) => void) | undefined;\n const claim = await startWatchClaim(configDir, opts.force ?? false, claimIntervalMs, (err) => {\n claimFailure ??= err;\n requestShutdown?.(err);\n });\n let claimCloseStarted = false;\n const closeClaim = (): Promise<void> => {\n claimCloseStarted = true;\n return claim.close();\n };\n try {\n const { store: initialStore, dbPath, warnings: initialWarnings, parsed: initialParsed } = await openStore(cfg);\n let initialTotal = 0;\n try {\n // total is read before close: an event fires only once nothing is held open again.\n initialTotal = initialWarnings.length > 0 || initialParsed > 0 ? await docCount(initialStore) : 0;\n } finally {\n await initialStore.close();\n }\n if (claimFailure) throw claimFailure;\n\n let stopping = false;\n // At most one open+reconcile+close cycle runs at a time; a trigger that arrives mid-cycle is\n // coalesced into a single rerun instead of opening a second, overlapping connection.\n let running: Promise<void> | null = null;\n let queuedAlways: boolean | null = null;\n\n const tick = async (alwaysEmit: boolean): Promise<void> => {\n try {\n const { store, parsed, warnings } = await openStore(cfg);\n try {\n if (alwaysEmit || parsed > 0 || warnings.length > 0) {\n onEvent({ type: 'reconciled', parsed, total: await docCount(store), warnings });\n }\n } finally {\n await store.close();\n }\n } catch (err) {\n onEvent({ type: 'reconcile-error', message: (err as Error).message });\n }\n };\n\n const startCycle = (alwaysEmit: boolean) => {\n running = tick(alwaysEmit).then(() => {\n running = null;\n if (queuedAlways !== null && !stopping) {\n const again = queuedAlways;\n queuedAlways = null;\n startCycle(again);\n }\n });\n };\n\n const requestCycle = (alwaysEmit: boolean) => {\n if (stopping) return;\n if (running) {\n queuedAlways = queuedAlways || alwaysEmit;\n return;\n }\n startCycle(alwaysEmit);\n };\n\n let debounceTimer: NodeJS.Timeout | null = null;\n const scheduleReconcile = () => {\n if (debounceTimer) clearTimeout(debounceTimer);\n debounceTimer = setTimeout(() => {\n debounceTimer = null;\n requestCycle(true);\n }, debounceMs);\n };\n\n // Ignore our own state files, or cache and claim writes would retrigger reconciliation forever.\n // An unresolvable filename reconciles because parsing nothing costs less than missing a real edit.\n const watcher = fsWatch(rootDir, { recursive: true }, (_event, filename) => {\n if (!shouldReconcileWatchNotification(filename)) return;\n scheduleReconcile();\n });\n const reconcileTimer = setInterval(\n guardedTick(\n async () => requestCycle(false),\n () => stopping\n ),\n reconcileIntervalMs\n );\n\n return await new Promise<void>((resolveShutdown, rejectShutdown) => {\n let shutdownReason: Error | null = null;\n const shutdown = (reason?: Error) => {\n shutdownReason ??= reason ?? null;\n if (stopping) return;\n stopping = true;\n const shutdownTask = (async () => {\n process.off('SIGINT', stopNormally);\n process.off('SIGTERM', stopNormally);\n opts.signal?.removeEventListener('abort', stopNormally);\n clearInterval(reconcileTimer);\n if (debounceTimer) clearTimeout(debounceTimer);\n watcher.close();\n await running;\n let finalError: unknown;\n try {\n // A normal stop performs one final freshness pass. Ownership loss drains existing work\n // but starts nothing new because the replacement watcher now owns that responsibility.\n if (!shutdownReason) {\n const { store: finalStore } = await openStore(cfg);\n await finalStore.close();\n }\n } catch (err) {\n finalError = err;\n }\n try {\n await closeClaim();\n } catch (err) {\n if (finalError) throw new AggregateError([finalError, err], 'watch shutdown and claim cleanup both failed');\n throw err;\n }\n if (finalError) throw finalError;\n if (shutdownReason) throw shutdownReason;\n })();\n shutdownTask.then(resolveShutdown, rejectShutdown);\n };\n const stopNormally = () => shutdown();\n requestShutdown = shutdown;\n process.once('SIGINT', stopNormally);\n process.once('SIGTERM', stopNormally);\n if (claimFailure) shutdown(claimFailure);\n else if (opts.signal?.aborted) shutdown();\n else {\n opts.signal?.addEventListener('abort', stopNormally, { once: true });\n try {\n onEvent({ type: 'started', rootDir, dbPath });\n if (!stopping && (initialWarnings.length > 0 || initialParsed > 0)) {\n onEvent({ type: 'reconciled', parsed: initialParsed, total: initialTotal, warnings: initialWarnings });\n }\n } catch (err) {\n shutdown(err instanceof Error ? err : new Error(String(err)));\n }\n }\n });\n } catch (err) {\n if (claimCloseStarted) throw err;\n try {\n await closeClaim();\n } catch (closeError) {\n throw new AggregateError([err, closeError], 'watch setup and claim cleanup both failed');\n }\n throw err;\n } finally {\n requestShutdown = undefined;\n }\n}\n"],"names":["runWatch","DEBOUNCE_MS","cfg","opts","onEvent","debounceMs","reconcileIntervalMs","claimIntervalMs","rootDir","configDir","claimFailure","requestShutdown","claim","claimCloseStarted","closeClaim","initialStore","dbPath","initialWarnings","initialParsed","initialTotal","stopping","running","queuedAlways","tick","startCycle","requestCycle","debounceTimer","scheduleReconcile","watcher","reconcileTimer","err","closeError","heartbeatIntervalMs","WATCH_HEARTBEAT_INTERVAL_MS","Number","isFinite","SenseError","Math","min","baseDir","startWatchClaim","force","close","openStore","store","warnings","parsed","length","docCount","alwaysEmit","type","total","message","then","again","clearTimeout","setTimeout","fsWatch","recursive","_event","filename","shouldReconcileWatchNotification","setInterval","guardedTick","Promise","resolveShutdown","rejectShutdown","shutdownReason","shutdown","reason","shutdownTask","finalError","finalStore","process","off","stopNormally","signal","removeEventListener","clearInterval","AggregateError","once","aborted","addEventListener","Error","String","undefined"],"mappings":";;;;+BAwBsBA;;;eAAAA;;;sBAxBW;wBAEN;6BACC;mCACqB;uBACb;4BACyB;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAE7D,uJAAuJ;AACvJ,IAAMC,cAAc;AAeb,SAAeD;wCAASE,GAAmB;YAAEC,MAClCA,eACGA,kBACSA,2BAIZD,cACEA,gBAG6BC,aAVzCC,SACAC,YACAC,qBAGAC,iBACAC,SACAC,WACFC,cACAC,iBACEC,OAIFC,mBACEC,YAKsF,MAA3EC,cAAcC,QAAkBC,iBAAyBC,eACpEC,oBASAC,UAGAC,SACAC,cAEEC,MAeAC,YAWAC,cASFC,eACEC,mBAUAC,SAIAC,gBA8DCC,KAIEC;;;;;oBAzJuC5B,OAAAA,oEAAqB,CAAC;oBAClEC,WAAUD,gBAAAA,KAAKC,OAAO,cAAZD,2BAAAA,gBAAiB,YAAO;oBAClCE,cAAaF,mBAAAA,KAAKE,UAAU,cAAfF,8BAAAA,mBAAmBF;oBAChCK,uBAAsBH,4BAAAA,KAAK6B,mBAAmB,cAAxB7B,uCAAAA,4BAA4B8B,yCAA2B;oBACnF,IAAI,CAACC,OAAOC,QAAQ,CAAC7B,wBAAwBA,uBAAuB,GAAG,MAAM,IAAI8B,oBAAU,CAAC,kBAAkB;oBAC9G,gGAAgG;oBAC1F7B,kBAAkB8B,KAAKC,GAAG,CAAChC,qBAAqB2B,yCAA2B;oBAC3EzB,WAAUN,eAAAA,IAAIM,OAAO,cAAXN,0BAAAA,eAAeA,IAAIqC,OAAO;oBACpC9B,aAAYP,iBAAAA,IAAIO,SAAS,cAAbP,4BAAAA,iBAAiBM;oBAC/BE,eAA6B;oBAEnB;;wBAAM8B,IAAAA,6BAAe,EAAC/B,YAAWN,cAAAA,KAAKsC,KAAK,cAAVtC,yBAAAA,cAAc,OAAOI,iBAAiB,SAACuB;4BACpFpB,yBAAAA,0BAAAA,eAAAA,eAAiBoB;4BACjBnB,4BAAAA,sCAAAA,gBAAkBmB;wBACpB;;;oBAHMlB,QAAQ;oBAIVC,oBAAoB;oBAClBC,aAAa;wBACjBD,oBAAoB;wBACpB,OAAOD,MAAM8B,KAAK;oBACpB;;;;;;;;;oBAE4F;;wBAAMC,IAAAA,kBAAS,EAACzC;;;oBAAhB,OAAA,eAA3Ea,eAA2E,KAAlF6B,OAAqB5B,SAA6D,KAA7DA,QAAkBC,kBAA2C,KAArD4B,UAAmC3B,gBAAkB,KAA1B4B;oBAC5D3B,eAAe;;;;;;;;;yBAGFF,CAAAA,gBAAgB8B,MAAM,GAAG,KAAK7B,gBAAgB,CAAA,GAA9CD;;;;oBAAkD;;wBAAM+B,IAAAA,iBAAQ,EAACjC;;;2BAAf;;;;;;2BAA+B;;;oBADhG,mFAAmF;oBACnFI;;;;;;oBAEA;;wBAAMJ,aAAa2B,KAAK;;;oBAAxB;;;;;oBAEF,IAAIhC,cAAc,MAAMA;oBAEpBU,WAAW;oBACf,6FAA6F;oBAC7F,qFAAqF;oBACjFC,UAAgC;oBAChCC,eAA+B;oBAE7BC,OAAO,cAAO0B;;gCAEoB,MAA5BL,OAAOE,QAAQD,gBAQhBf;;;;;;;;;;wCAR6B;;4CAAMa,IAAAA,kBAAS,EAACzC;;;wCAAhB,OAAA,eAA5B0C,QAA4B,KAA5BA,OAAOE,SAAqB,KAArBA,QAAQD,WAAa,KAAbA;;;;;;;;;6CAEjBI,CAAAA,cAAcH,SAAS,KAAKD,SAASE,MAAM,GAAG,CAAA,GAA9CE;;;;;4CACQC,MAAM;4CAAcJ,QAAAA;;wCAAe;;4CAAME,IAAAA,iBAAQ,EAACJ;;;wCAA5DxC;kDAAsC+C,QAAO,oBAAuBN,WAAAA;;;;;;;;;wCAGtE;;4CAAMD,MAAMF,KAAK;;;wCAAjB;;;;;;;;;;wCAEKZ;wCACP1B,QAAQ;4CAAE8C,MAAM;4CAAmBE,SAAS,AAACtB,IAAcsB,OAAO;wCAAC;;;;;;;;;;;wBAEvE;;oBAEM5B,aAAa,qBAACyB;wBAClB5B,UAAUE,KAAK0B,YAAYI,IAAI,CAAC;4BAC9BhC,UAAU;4BACV,IAAIC,iBAAiB,QAAQ,CAACF,UAAU;gCACtC,IAAMkC,QAAQhC;gCACdA,eAAe;gCACfE,WAAW8B;4BACb;wBACF;oBACF;oBAEM7B,eAAe,sBAACwB;wBACpB,IAAI7B,UAAU;wBACd,IAAIC,SAAS;4BACXC,eAAeA,gBAAgB2B;4BAC/B;wBACF;wBACAzB,WAAWyB;oBACb;oBAEIvB,gBAAuC;oBACrCC,oBAAoB;wBACxB,IAAID,eAAe6B,aAAa7B;wBAChCA,gBAAgB8B,WAAW;4BACzB9B,gBAAgB;4BAChBD,aAAa;wBACf,GAAGpB;oBACL;oBAEA,gGAAgG;oBAChG,mGAAmG;oBAC7FuB,UAAU6B,IAAAA,aAAO,EAACjD,SAAS;wBAAEkD,WAAW;oBAAK,GAAG,SAACC,QAAQC;wBAC7D,IAAI,CAACC,IAAAA,qDAAgC,EAACD,WAAW;wBACjDjC;oBACF;oBACME,iBAAiBiC,YACrBC,IAAAA,0BAAW,EACT;;;;;oCAAYtC,aAAa;;;;uBACzB;+BAAML;wBAERd;oBAGK;;wBAAM,IAAI0D,QAAc,SAACC,iBAAiBC;gCAyCtC/D;4BAxCT,IAAIgE,iBAA+B;4BACnC,IAAMC,WAAW,kBAACC;gCAChBF,2BAAAA,4BAAAA,iBAAAA,iBAAmBE,mBAAAA,oBAAAA,SAAU;gCAC7B,IAAIjD,UAAU;gCACdA,WAAW;gCACX,IAAMkD,eAAe,AAAC;;4CAGpBnE,cAKIoE,YAK8B,MAAfC,YAGV1C,KAKAA;;;;oDApBT2C,QAAQC,GAAG,CAAC,UAAUC;oDACtBF,QAAQC,GAAG,CAAC,WAAWC;qDACvBxE,eAAAA,KAAKyE,MAAM,cAAXzE,mCAAAA,aAAa0E,mBAAmB,CAAC,SAASF;oDAC1CG,cAAcjD;oDACd,IAAIH,eAAe6B,aAAa7B;oDAChCE,QAAQc,KAAK;oDACb;;wDAAMrB;;;oDAAN;;;;;;;;;yDAKM,CAAC8C,gBAAD;;;;oDAC4B;;wDAAMxB,IAAAA,kBAAS,EAACzC;;;oDAAhB,OAAA,eAAfsE,aAAe,KAAtB5B;oDACR;;wDAAM4B,WAAW9B,KAAK;;;oDAAtB;;;;;;;;oDAEKZ;oDACPyC,aAAazC;;;;;;;;;;;;oDAGb;;wDAAMhB;;;oDAAN;;;;;;oDACOgB;oDACP,IAAIyC,YAAY,MAAM,IAAIQ;wDAAgBR;wDAAYzC;uDAAM;oDAC5D,MAAMA;;oDAER,IAAIyC,YAAY,MAAMA;oDACtB,IAAIJ,gBAAgB,MAAMA;;;;;;oCAC5B;;gCACAG,aAAajB,IAAI,CAACY,iBAAiBC;4BACrC;4BACA,IAAMS,eAAe;uCAAMP;;4BAC3BzD,kBAAkByD;4BAClBK,QAAQO,IAAI,CAAC,UAAUL;4BACvBF,QAAQO,IAAI,CAAC,WAAWL;4BACxB,IAAIjE,cAAc0D,SAAS1D;iCACtB,KAAIP,eAAAA,KAAKyE,MAAM,cAAXzE,mCAAAA,aAAa8E,OAAO,EAAEb;iCAC1B;oCACHjE;iCAAAA,gBAAAA,KAAKyE,MAAM,cAAXzE,oCAAAA,cAAa+E,gBAAgB,CAAC,SAASP,cAAc;oCAAEK,MAAM;gCAAK;gCAClE,IAAI;oCACF5E,QAAQ;wCAAE8C,MAAM;wCAAW1C,SAAAA;wCAASQ,QAAAA;oCAAO;oCAC3C,IAAI,CAACI,YAAaH,CAAAA,gBAAgB8B,MAAM,GAAG,KAAK7B,gBAAgB,CAAA,GAAI;wCAClEd,QAAQ;4CAAE8C,MAAM;4CAAcJ,QAAQ5B;4CAAeiC,OAAOhC;4CAAc0B,UAAU5B;wCAAgB;oCACtG;gCACF,EAAE,OAAOa,KAAK;oCACZsC,SAAStC,AAAG,YAAHA,KAAeqD,SAAQrD,MAAM,IAAIqD,MAAMC,OAAOtD;gCACzD;4BACF;wBACF;;;oBArDA;;wBAAO;;;oBAsDAA;oBACP,IAAIjB,mBAAmB,MAAMiB;;;;;;;;;oBAE3B;;wBAAMhB;;;oBAAN;;;;;;oBACOiB;oBACP,MAAM,IAAIgD;wBAAgBjD;wBAAKC;uBAAa;;oBAE9C,MAAMD;;oBAENnB,kBAAkB0E;;;;;;;;;;IAEtB"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
export declare function shouldReconcileWatchNotification(filename: string | Buffer | null): boolean;
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/lib/watch-notification.ts"],"sourcesContent":["import { STATE_DIR } from '../config/types.ts';\n\nexport function shouldReconcileWatchNotification(filename: string | Buffer | null): boolean {\n return typeof filename !== 'string' || !filename.startsWith(STATE_DIR);\n}\n"],"names":["STATE_DIR","shouldReconcileWatchNotification","filename","startsWith"],"mappings":"AAAA,SAASA,SAAS,QAAQ,qBAAqB;AAE/C,OAAO,SAASC,iCAAiCC,QAAgC;IAC/E,OAAO,OAAOA,aAAa,YAAY,CAACA,SAASC,UAAU,CAACH;AAC9D"}
|
|
@@ -1,4 +1,4 @@
|
|
|
1
1
|
export declare const DUCKDB_PACKAGE = "@duckdb/node-api";
|
|
2
|
-
export declare const DUCKDB_VERSION = "1.5.5-r.
|
|
3
|
-
export declare const DUCKDB_INSTALL_SPEC = "@duckdb/node-api@1.5.5-r.
|
|
2
|
+
export declare const DUCKDB_VERSION = "1.5.5-r.5";
|
|
3
|
+
export declare const DUCKDB_INSTALL_SPEC = "@duckdb/node-api@1.5.5-r.5";
|
|
4
4
|
export declare function duckdbApi(): Promise<typeof import('@duckdb/node-api')>;
|
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
import { loadOrInstall, packageNodeModules } from '../native.js';
|
|
2
2
|
export const DUCKDB_PACKAGE = '@duckdb/node-api';
|
|
3
|
-
export const DUCKDB_VERSION = '1.5.5-r.
|
|
3
|
+
export const DUCKDB_VERSION = '1.5.5-r.5';
|
|
4
4
|
export const DUCKDB_INSTALL_SPEC = `${DUCKDB_PACKAGE}@${DUCKDB_VERSION}`;
|
|
5
5
|
const DUCKDB = {
|
|
6
6
|
store: 'duckdb',
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/store/duckdb/native.ts"],"sourcesContent":["import { loadOrInstall, type NativeDescriptor, packageNodeModules } from '../native.ts';\n\nexport const DUCKDB_PACKAGE = '@duckdb/node-api';\nexport const DUCKDB_VERSION = '1.5.5-r.
|
|
1
|
+
{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/store/duckdb/native.ts"],"sourcesContent":["import { loadOrInstall, type NativeDescriptor, packageNodeModules } from '../native.ts';\n\nexport const DUCKDB_PACKAGE = '@duckdb/node-api';\nexport const DUCKDB_VERSION = '1.5.5-r.5';\nexport const DUCKDB_INSTALL_SPEC = `${DUCKDB_PACKAGE}@${DUCKDB_VERSION}`;\n\nconst DUCKDB: NativeDescriptor = {\n store: 'duckdb',\n pkg: DUCKDB_PACKAGE,\n sizeHint: '~110MB',\n version: DUCKDB_VERSION,\n installSpec: DUCKDB_INSTALL_SPEC,\n};\n\nlet duckdbApiPromise: Promise<typeof import('@duckdb/node-api')> | undefined;\n\n// The one accessor every consumer of @duckdb/node-api must go through: after a fresh install in this process, Node keeps returning\n// the pre-install \"not found\" miss to any new bare-specifier lookup of the same path, so caching the resolved module here means only the first caller ever resolves it.\nexport function duckdbApi(): Promise<typeof import('@duckdb/node-api')> {\n if (!duckdbApiPromise) duckdbApiPromise = loadOrInstall<typeof import('@duckdb/node-api')>(DUCKDB, packageNodeModules());\n return duckdbApiPromise;\n}\n"],"names":["loadOrInstall","packageNodeModules","DUCKDB_PACKAGE","DUCKDB_VERSION","DUCKDB_INSTALL_SPEC","DUCKDB","store","pkg","sizeHint","version","installSpec","duckdbApiPromise","duckdbApi"],"mappings":"AAAA,SAASA,aAAa,EAAyBC,kBAAkB,QAAQ,eAAe;AAExF,OAAO,MAAMC,iBAAiB,mBAAmB;AACjD,OAAO,MAAMC,iBAAiB,YAAY;AAC1C,OAAO,MAAMC,sBAAsB,GAAGF,eAAe,CAAC,EAAEC,gBAAgB,CAAC;AAEzE,MAAME,SAA2B;IAC/BC,OAAO;IACPC,KAAKL;IACLM,UAAU;IACVC,SAASN;IACTO,aAAaN;AACf;AAEA,IAAIO;AAEJ,mIAAmI;AACnI,wKAAwK;AACxK,OAAO,SAASC;IACd,IAAI,CAACD,kBAAkBA,mBAAmBX,cAAiDK,QAAQJ;IACnG,OAAOU;AACT"}
|
|
@@ -9,13 +9,14 @@ import { DUCKDB_PACKAGE, duckdbApi } from './native.js';
|
|
|
9
9
|
import { duckdbDialect } from './reconcile.js';
|
|
10
10
|
import { registerFunctions } from './sql-functions.js';
|
|
11
11
|
import { createStore } from './store.js';
|
|
12
|
+
import { createVectorWriteStage } from './vectors.js';
|
|
12
13
|
export const DB_FILENAME = 'cache.duckdb';
|
|
13
14
|
// Independent of sqlite's SCHEMA_VERSION -- the two stores' cache shapes evolve separately (VARIANT frontmatter columns vs untyped).
|
|
14
15
|
// The store name already joins the feature signature, so switching a config's `store` key rebuilds rather than reusing the other engine's cache.
|
|
15
16
|
export const SCHEMA_VERSION = '4';
|
|
16
17
|
// `content` is a plain table (not FTS-virtual): the fts index is built over it lazily, only when a lexical query runs (lexical.ts), and read directly for contains() verification either way.
|
|
17
18
|
// No tokenizer resolution: this store always uses the fts extension's default (porter) stemmer.
|
|
18
|
-
async function ensureSchema(
|
|
19
|
+
async function ensureSchema(handle, conn, cfg) {
|
|
19
20
|
await conn.exec(`CREATE TABLE IF NOT EXISTS frontmatter ("path" TEXT PRIMARY KEY, "_mtime" DOUBLE, "_ctime" DOUBLE, "_size" INTEGER, "_parse_error" TEXT)`);
|
|
20
21
|
await conn.exec(`CREATE TABLE IF NOT EXISTS content ("path" TEXT PRIMARY KEY, title TEXT, summary TEXT, text TEXT)`);
|
|
21
22
|
await conn.exec(`CREATE TABLE IF NOT EXISTS preset_files ("path" TEXT, preset TEXT, PRIMARY KEY ("path", preset))`);
|
|
@@ -25,6 +26,7 @@ async function ensureSchema(_handle, conn, cfg) {
|
|
|
25
26
|
// so the feature's shared reconcile-time INSERT/DELETE (features/embed.ts) names a column that exists on both stores.
|
|
26
27
|
if (feature.name === 'embed') {
|
|
27
28
|
await conn.exec(`CREATE TABLE IF NOT EXISTS embeddings ("path" TEXT, chunk INTEGER, start_line INTEGER, end_line INTEGER, scale REAL, vector FLOAT[${STORE_DIMS}], PRIMARY KEY ("path", chunk))`);
|
|
29
|
+
await createVectorWriteStage(handle.duckdb, STORE_DIMS);
|
|
28
30
|
continue;
|
|
29
31
|
}
|
|
30
32
|
await feature.schema(conn);
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/store/duckdb/open.ts"],"sourcesContent":["import type { DuckDBConnection, DuckDBInstance } from '@duckdb/node-api';\nimport type { Config, ResolvedConfig } from '../../config/index.ts';\nimport { featureSignature } from '../../config/index.ts';\nimport { STORE_DIMS } from '../../embed/types.ts';\nimport { SenseError } from '../../errors.ts';\nimport { activeFeatures, FEATURES } from '../../features/index.ts';\nimport type { OpenResult } from '../open.ts';\nimport { openWithDialect } from '../open.ts';\nimport { getMeta, setMeta } from '../shared.ts';\nimport type { Connection, OpenDialect } from '../types.ts';\nimport { createConnection } from './connection.ts';\nimport { DUCKDB_PACKAGE, duckdbApi } from './native.ts';\nimport { duckdbDialect } from './reconcile.ts';\nimport { registerFunctions } from './sql-functions.ts';\nimport { createStore } from './store.ts';\n\nexport const DB_FILENAME = 'cache.duckdb';\n// Independent of sqlite's SCHEMA_VERSION -- the two stores' cache shapes evolve separately (VARIANT frontmatter columns vs untyped).\n// The store name already joins the feature signature, so switching a config's `store` key rebuilds rather than reusing the other engine's cache.\nexport const SCHEMA_VERSION = '4';\n\nexport type { OpenResult };\n\n// The connection plus this store's own native handles (types.ts's OpenDialect<Handle>): the\n// instance owns the WAL and must outlive the connection borrowed from it.\ninterface DuckdbHandle {\n instance: DuckDBInstance;\n duckdb: DuckDBConnection;\n}\n\n// `content` is a plain table (not FTS-virtual): the fts index is built over it lazily, only when a lexical query runs (lexical.ts), and read directly for contains() verification either way.\n// No tokenizer resolution: this store always uses the fts extension's default (porter) stemmer.\nasync function ensureSchema(
|
|
1
|
+
{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/store/duckdb/open.ts"],"sourcesContent":["import type { DuckDBConnection, DuckDBInstance } from '@duckdb/node-api';\nimport type { Config, ResolvedConfig } from '../../config/index.ts';\nimport { featureSignature } from '../../config/index.ts';\nimport { STORE_DIMS } from '../../embed/types.ts';\nimport { SenseError } from '../../errors.ts';\nimport { activeFeatures, FEATURES } from '../../features/index.ts';\nimport type { OpenResult } from '../open.ts';\nimport { openWithDialect } from '../open.ts';\nimport { getMeta, setMeta } from '../shared.ts';\nimport type { Connection, OpenDialect } from '../types.ts';\nimport { createConnection } from './connection.ts';\nimport { DUCKDB_PACKAGE, duckdbApi } from './native.ts';\nimport { duckdbDialect } from './reconcile.ts';\nimport { registerFunctions } from './sql-functions.ts';\nimport { createStore } from './store.ts';\nimport { createVectorWriteStage } from './vectors.ts';\n\nexport const DB_FILENAME = 'cache.duckdb';\n// Independent of sqlite's SCHEMA_VERSION -- the two stores' cache shapes evolve separately (VARIANT frontmatter columns vs untyped).\n// The store name already joins the feature signature, so switching a config's `store` key rebuilds rather than reusing the other engine's cache.\nexport const SCHEMA_VERSION = '4';\n\nexport type { OpenResult };\n\n// The connection plus this store's own native handles (types.ts's OpenDialect<Handle>): the\n// instance owns the WAL and must outlive the connection borrowed from it.\ninterface DuckdbHandle {\n instance: DuckDBInstance;\n duckdb: DuckDBConnection;\n}\n\n// `content` is a plain table (not FTS-virtual): the fts index is built over it lazily, only when a lexical query runs (lexical.ts), and read directly for contains() verification either way.\n// No tokenizer resolution: this store always uses the fts extension's default (porter) stemmer.\nasync function ensureSchema(handle: DuckdbHandle, conn: Connection, cfg: Config): Promise<void> {\n await conn.exec(`CREATE TABLE IF NOT EXISTS frontmatter (\"path\" TEXT PRIMARY KEY, \"_mtime\" DOUBLE, \"_ctime\" DOUBLE, \"_size\" INTEGER, \"_parse_error\" TEXT)`);\n await conn.exec(`CREATE TABLE IF NOT EXISTS content (\"path\" TEXT PRIMARY KEY, title TEXT, summary TEXT, text TEXT)`);\n await conn.exec(`CREATE TABLE IF NOT EXISTS preset_files (\"path\" TEXT, preset TEXT, PRIMARY KEY (\"path\", preset))`);\n await conn.exec('CREATE INDEX IF NOT EXISTS preset_files_preset ON preset_files(preset)');\n for (const feature of activeFeatures(cfg)) {\n // Native FLOAT[STORE_DIMS] instead of the embed feature's engine-neutral BLOB+scale DDL (vectors.ts). `scale` is kept, unused,\n // so the feature's shared reconcile-time INSERT/DELETE (features/embed.ts) names a column that exists on both stores.\n if (feature.name === 'embed') {\n await conn.exec(`CREATE TABLE IF NOT EXISTS embeddings (\"path\" TEXT, chunk INTEGER, start_line INTEGER, end_line INTEGER, scale REAL, vector FLOAT[${STORE_DIMS}], PRIMARY KEY (\"path\", chunk))`);\n await createVectorWriteStage(handle.duckdb, STORE_DIMS);\n continue;\n }\n await feature.schema(conn);\n }\n if ((await getMeta(conn, 'schema_version')) === null) await setMeta(conn, 'schema_version', SCHEMA_VERSION);\n if ((await getMeta(conn, 'features')) === null) await setMeta(conn, 'features', featureSignature(cfg, FEATURES));\n}\n\n// Order matters: the connection must be gone before the instance closes the WAL.\nasync function close(handle: DuckdbHandle): Promise<void> {\n handle.duckdb.disconnectSync();\n handle.instance.closeSync();\n}\n\nasync function connect(dbPath: string, _cfg: ResolvedConfig): Promise<{ handle: DuckdbHandle; conn: Connection }> {\n // Dynamic, not a top-level import: sqlite trees must never attempt to resolve this optional peer dependency, so nothing imports\n // it as a value until a duckdb tree opens (types-only imports are erased). Installed on first use if missing, shared with sql-functions.ts and vectors.ts via native.ts's duckdbApi.\n let DuckDBInstance: typeof import('@duckdb/node-api').DuckDBInstance;\n try {\n ({ DuckDBInstance } = await duckdbApi());\n } catch (err) {\n if (err instanceof SenseError) throw err;\n throw new SenseError('STORE_DEPENDENCY_MISSING', `store \"duckdb\" needs the ${DUCKDB_PACKAGE} package (${(err as Error).message})`);\n }\n\n let duckdb: DuckDBConnection;\n let instance: DuckDBInstance | undefined;\n try {\n // On-disk files default to an older storage format for cross-version compatibility, which rejects VARIANT columns (\"VARIANT\n // columns are not supported in storage versions prior to v1.5.0\"); this store's dynamic frontmatter columns need VARIANT (reconcile.ts), so the floor is pinned explicitly.\n instance = await DuckDBInstance.create(dbPath, { storage_compatibility_version: 'v1.5.0' });\n duckdb = await instance.connect();\n } catch (err) {\n // create() may have succeeded before connect() failed: close it, or its WAL stays open.\n instance?.closeSync();\n throw new SenseError('STORE_DEPENDENCY_MISSING', `store \"duckdb\" failed to open ${dbPath}: ${(err as Error).message}`);\n }\n // A throw below would leak the open instance, whose WAL then locks the .duckdb file undeletable on Windows.\n try {\n await registerFunctions(duckdb);\n const conn = createConnection(duckdb);\n return { handle: { instance, duckdb }, conn };\n } catch (err) {\n await close({ instance, duckdb });\n throw err;\n }\n}\n\n// This store's dialect (types.ts's OpenDialect) for the shared orchestration in store/open.ts.\nexport const duckdbOpenDialect: OpenDialect<DuckdbHandle> = {\n filename: DB_FILENAME,\n schemaVersion: SCHEMA_VERSION,\n reconcileDialect: duckdbDialect,\n connect,\n close,\n // duckdb's file lock spans the connection's life, and each platform words the refusal its own\n // way: posix \"Could not set lock on file ... Conflicting lock is held in <exe> (PID n)\", Windows\n // \"Cannot open file ... being used by another process\". Both are the same condition.\n isLocked: (err) => /Could not set lock on file|being used by another process/.test(err.message),\n ensureSchema,\n createStore: (handle, conn) => createStore(handle.instance, handle.duckdb, conn),\n};\n\nexport async function openDuckdb(cfg: ResolvedConfig): Promise<OpenResult> {\n return openWithDialect(cfg, duckdbOpenDialect);\n}\n"],"names":["featureSignature","STORE_DIMS","SenseError","activeFeatures","FEATURES","openWithDialect","getMeta","setMeta","createConnection","DUCKDB_PACKAGE","duckdbApi","duckdbDialect","registerFunctions","createStore","createVectorWriteStage","DB_FILENAME","SCHEMA_VERSION","ensureSchema","handle","conn","cfg","exec","feature","name","duckdb","schema","close","disconnectSync","instance","closeSync","connect","dbPath","_cfg","DuckDBInstance","err","message","create","storage_compatibility_version","duckdbOpenDialect","filename","schemaVersion","reconcileDialect","isLocked","test","openDuckdb"],"mappings":"AAEA,SAASA,gBAAgB,QAAQ,wBAAwB;AACzD,SAASC,UAAU,QAAQ,uBAAuB;AAClD,SAASC,UAAU,QAAQ,kBAAkB;AAC7C,SAASC,cAAc,EAAEC,QAAQ,QAAQ,0BAA0B;AAEnE,SAASC,eAAe,QAAQ,aAAa;AAC7C,SAASC,OAAO,EAAEC,OAAO,QAAQ,eAAe;AAEhD,SAASC,gBAAgB,QAAQ,kBAAkB;AACnD,SAASC,cAAc,EAAEC,SAAS,QAAQ,cAAc;AACxD,SAASC,aAAa,QAAQ,iBAAiB;AAC/C,SAASC,iBAAiB,QAAQ,qBAAqB;AACvD,SAASC,WAAW,QAAQ,aAAa;AACzC,SAASC,sBAAsB,QAAQ,eAAe;AAEtD,OAAO,MAAMC,cAAc,eAAe;AAC1C,qIAAqI;AACrI,iJAAiJ;AACjJ,OAAO,MAAMC,iBAAiB,IAAI;AAWlC,8LAA8L;AAC9L,gGAAgG;AAChG,eAAeC,aAAaC,MAAoB,EAAEC,IAAgB,EAAEC,GAAW;IAC7E,MAAMD,KAAKE,IAAI,CAAC,CAAC,wIAAwI,CAAC;IAC1J,MAAMF,KAAKE,IAAI,CAAC,CAAC,iGAAiG,CAAC;IACnH,MAAMF,KAAKE,IAAI,CAAC,CAAC,gGAAgG,CAAC;IAClH,MAAMF,KAAKE,IAAI,CAAC;IAChB,KAAK,MAAMC,WAAWnB,eAAeiB,KAAM;QACzC,+HAA+H;QAC/H,sHAAsH;QACtH,IAAIE,QAAQC,IAAI,KAAK,SAAS;YAC5B,MAAMJ,KAAKE,IAAI,CAAC,CAAC,kIAAkI,EAAEpB,WAAW,+BAA+B,CAAC;YAChM,MAAMa,uBAAuBI,OAAOM,MAAM,EAAEvB;YAC5C;QACF;QACA,MAAMqB,QAAQG,MAAM,CAACN;IACvB;IACA,IAAI,AAAC,MAAMb,QAAQa,MAAM,sBAAuB,MAAM,MAAMZ,QAAQY,MAAM,kBAAkBH;IAC5F,IAAI,AAAC,MAAMV,QAAQa,MAAM,gBAAiB,MAAM,MAAMZ,QAAQY,MAAM,YAAYnB,iBAAiBoB,KAAKhB;AACxG;AAEA,iFAAiF;AACjF,eAAesB,MAAMR,MAAoB;IACvCA,OAAOM,MAAM,CAACG,cAAc;IAC5BT,OAAOU,QAAQ,CAACC,SAAS;AAC3B;AAEA,eAAeC,QAAQC,MAAc,EAAEC,IAAoB;IACzD,gIAAgI;IAChI,qLAAqL;IACrL,IAAIC;IACJ,IAAI;QACD,CAAA,EAAEA,cAAc,EAAE,GAAG,MAAMvB,WAAU;IACxC,EAAE,OAAOwB,KAAK;QACZ,IAAIA,eAAehC,YAAY,MAAMgC;QACrC,MAAM,IAAIhC,WAAW,4BAA4B,CAAC,yBAAyB,EAAEO,eAAe,UAAU,EAAE,AAACyB,IAAcC,OAAO,CAAC,CAAC,CAAC;IACnI;IAEA,IAAIX;IACJ,IAAII;IACJ,IAAI;QACF,4HAA4H;QAC5H,4KAA4K;QAC5KA,WAAW,MAAMK,eAAeG,MAAM,CAACL,QAAQ;YAAEM,+BAA+B;QAAS;QACzFb,SAAS,MAAMI,SAASE,OAAO;IACjC,EAAE,OAAOI,KAAK;QACZ,wFAAwF;QACxFN,qBAAAA,+BAAAA,SAAUC,SAAS;QACnB,MAAM,IAAI3B,WAAW,4BAA4B,CAAC,8BAA8B,EAAE6B,OAAO,EAAE,EAAE,AAACG,IAAcC,OAAO,EAAE;IACvH;IACA,4GAA4G;IAC5G,IAAI;QACF,MAAMvB,kBAAkBY;QACxB,MAAML,OAAOX,iBAAiBgB;QAC9B,OAAO;YAAEN,QAAQ;gBAAEU;gBAAUJ;YAAO;YAAGL;QAAK;IAC9C,EAAE,OAAOe,KAAK;QACZ,MAAMR,MAAM;YAAEE;YAAUJ;QAAO;QAC/B,MAAMU;IACR;AACF;AAEA,+FAA+F;AAC/F,OAAO,MAAMI,oBAA+C;IAC1DC,UAAUxB;IACVyB,eAAexB;IACfyB,kBAAkB9B;IAClBmB;IACAJ;IACA,8FAA8F;IAC9F,iGAAiG;IACjG,qFAAqF;IACrFgB,UAAU,CAACR,MAAQ,2DAA2DS,IAAI,CAACT,IAAIC,OAAO;IAC9FlB;IACAJ,aAAa,CAACK,QAAQC,OAASN,YAAYK,OAAOU,QAAQ,EAAEV,OAAOM,MAAM,EAAEL;AAC7E,EAAE;AAEF,OAAO,eAAeyB,WAAWxB,GAAmB;IAClD,OAAOf,gBAAgBe,KAAKkB;AAC9B"}
|
|
@@ -1,5 +1,6 @@
|
|
|
1
1
|
import type { DuckDBConnection } from '@duckdb/node-api';
|
|
2
2
|
import type { Connection, VectorCandidate, VectorSimilar, VectorWriteRow } from '../types.js';
|
|
3
|
+
export declare function createVectorWriteStage(duckdb: DuckDBConnection, dims: number): Promise<void>;
|
|
3
4
|
export declare function writeVectorBatch(duckdb: DuckDBConnection, conn: Connection, dims: number, rows: VectorWriteRow[]): Promise<void>;
|
|
4
5
|
export declare function scanCandidates(duckdb: DuckDBConnection, qv: Float32Array, dims: number, fetch: number, allowed?: Set<string>): Promise<VectorCandidate[]>;
|
|
5
6
|
export declare function scanSimilar(duckdb: DuckDBConnection, conn: Connection, dims: number, path: string, opts: {
|
|
@@ -1,6 +1,5 @@
|
|
|
1
1
|
import { withTransaction } from '../transaction.js';
|
|
2
2
|
import { asCosine, sampleEvenly } from '../vectors.js';
|
|
3
|
-
import { rewriteUpdate } from './batch.js';
|
|
4
3
|
import { duckdbApi } from './native.js';
|
|
5
4
|
// This store keeps vectors as native FLOAT[dims] arrays (dims fixed at DDL time by open.ts's ensureSchema, from embed/types.ts's STORE_DIMS) and
|
|
6
5
|
// scans them with array_cosine_similarity, pushing top-k into SQL instead of pulling every row into JS; sqlite's int8+scale BLOB scan is its own representation. Every function here takes `dims` as a parameter rather than importing STORE_DIMS directly, so it stays testable at any width.
|
|
@@ -9,6 +8,11 @@ import { duckdbApi } from './native.js';
|
|
|
9
8
|
// A bind position whose type is left to auto-inference (safe for plain strings/numbers; only
|
|
10
9
|
// the vector ARRAY positions below need an explicit type -- see writeVectorBatch's comment).
|
|
11
10
|
const untyped = undefined;
|
|
11
|
+
const VECTOR_WRITE_STAGE = '_sense_vector_write_stage';
|
|
12
|
+
// Created once per native connection outside write transactions; its lifetime follows that connection.
|
|
13
|
+
export async function createVectorWriteStage(duckdb, dims) {
|
|
14
|
+
await duckdb.run(`CREATE TEMP TABLE ${VECTOR_WRITE_STAGE} ("path" TEXT, chunk INTEGER, vector FLOAT[${dims}])`);
|
|
15
|
+
}
|
|
12
16
|
// The DDL-fixed array width can exceed a vector's actual length (a hypothetical model sliced under the column's width);
|
|
13
17
|
// zero-padding leaves cosine scores unchanged since the added dimensions contribute nothing to either vector's dot product or norm.
|
|
14
18
|
function padded(values, dims) {
|
|
@@ -27,27 +31,27 @@ function dequantize(row, dims) {
|
|
|
27
31
|
const q = new Int8Array(row.vector.buffer, row.vector.byteOffset, row.vector.byteLength);
|
|
28
32
|
return padded(Array.from(q, (v)=>v * row.scale), dims);
|
|
29
33
|
}
|
|
30
|
-
//
|
|
31
|
-
//
|
|
34
|
+
// The native appender needs an explicit FLOAT array type: inference from an integer first component
|
|
35
|
+
// silently truncates fractional components. The fixed UPDATE keeps each provider batch keyed by path and chunk.
|
|
32
36
|
export async function writeVectorBatch(duckdb, conn, dims, rows) {
|
|
33
37
|
if (rows.length === 0) return;
|
|
34
|
-
const {
|
|
35
|
-
const
|
|
36
|
-
if (!rewritten) throw new Error('duckdb: writeVectorBatch SQL shape not recognized by rewriteUpdate');
|
|
38
|
+
const { ARRAY, FLOAT } = await duckdbApi();
|
|
39
|
+
const vectorType = ARRAY(FLOAT, dims);
|
|
37
40
|
await withTransaction(conn, async ()=>{
|
|
38
|
-
|
|
41
|
+
await duckdb.run(`DELETE FROM ${VECTOR_WRITE_STAGE}`);
|
|
42
|
+
const appender = await duckdb.createAppender(VECTOR_WRITE_STAGE);
|
|
39
43
|
try {
|
|
40
|
-
const values = [];
|
|
41
|
-
const types = [];
|
|
42
44
|
for (const row of rows){
|
|
43
|
-
|
|
44
|
-
|
|
45
|
+
appender.appendVarchar(row.path);
|
|
46
|
+
appender.appendInteger(row.chunk);
|
|
47
|
+
appender.appendArray(dequantize(row, dims), vectorType);
|
|
48
|
+
appender.endRow();
|
|
45
49
|
}
|
|
46
|
-
|
|
47
|
-
await stmt.run();
|
|
50
|
+
appender.flushSync();
|
|
48
51
|
} finally{
|
|
49
|
-
|
|
52
|
+
appender.closeSync();
|
|
50
53
|
}
|
|
54
|
+
await duckdb.run(`UPDATE embeddings SET vector = staged.vector FROM ${VECTOR_WRITE_STAGE} AS staged WHERE embeddings."path" = staged."path" AND embeddings.chunk = staged.chunk`);
|
|
51
55
|
});
|
|
52
56
|
}
|
|
53
57
|
// Best chunk per file by cosine, with the earliest chunk winning exact score ties.
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/store/duckdb/vectors.ts"],"sourcesContent":["import type { DuckDBConnection, DuckDBType, DuckDBValue } from '@duckdb/node-api';\nimport { withTransaction } from '../transaction.ts';\nimport type { Connection, VectorCandidate, VectorSimilar, VectorWriteRow } from '../types.ts';\nimport { asCosine, sampleEvenly } from '../vectors.ts';\nimport { rewriteUpdate } from './batch.ts';\nimport { duckdbApi } from './native.ts';\n\n// This store keeps vectors as native FLOAT[dims] arrays (dims fixed at DDL time by open.ts's ensureSchema, from embed/types.ts's STORE_DIMS) and\n// scans them with array_cosine_similarity, pushing top-k into SQL instead of pulling every row into JS; sqlite's int8+scale BLOB scan is its own representation. Every function here takes `dims` as a parameter rather than importing STORE_DIMS directly, so it stays testable at any width.\n\n// A value import of '@duckdb/node-api' must never sit at module top level, or a sqlite-only tree would resolve this optional dependency\n// just by importing store/index.ts; native.ts's shared duckdbApi() keeps that lazy and reuses whatever open.ts already resolved.\n\n// A bind position whose type is left to auto-inference (safe for plain strings/numbers; only\n// the vector ARRAY positions below need an explicit type -- see writeVectorBatch's comment).\nconst untyped = undefined as unknown as DuckDBType;\n\n// The DDL-fixed array width can exceed a vector's actual length (a hypothetical model sliced under the column's width);\n// zero-padding leaves cosine scores unchanged since the added dimensions contribute nothing to either vector's dot product or norm.\nfunction padded(values: ArrayLike<number>, dims: number): number[] {\n const out = new Array<number>(dims).fill(0);\n for (let d = 0; d < Math.min(values.length, dims); d++) out[d] = values[d];\n return out;\n}\n\nfunction inClause(column: string, count: number): string {\n return `${column} IN (${Array.from({ length: count }, () => '?').join(', ')})`;\n}\n\n// Rows arrive int8-quantized with a per-vector scale (the wire format embed/query.ts produces for every store, see toStore);\n// dequantizing into a plain float array is this store's own representation choice, so both stores are handed the same vectors and diverge only here.\nfunction dequantize(row: VectorWriteRow, dims: number): number[] {\n const q = new Int8Array(row.vector.buffer, row.vector.byteOffset, row.vector.byteLength);\n return padded(\n Array.from(q, (v) => v * row.scale),\n dims\n );\n}\n\n// One runBatch-equivalent call per provider batch: a single multi-row UPDATE (batch.ts's rewriteUpdate), bound directly against the\n// native connection because the vector column needs an explicit ARRAY(DOUBLE, dims) bind type -- auto inference misreads an all-integer first component as INTEGER and silently truncates the rest (verified live: an untyped bind of [0, 0.35, 0.1, -0.2] stores [0,0,0,0]).\nexport async function writeVectorBatch(duckdb: DuckDBConnection, conn: Connection, dims: number, rows: VectorWriteRow[]): Promise<void> {\n if (rows.length === 0) return;\n const { arrayValue, ARRAY, DOUBLE } = await duckdbApi();\n const rewritten = rewriteUpdate('UPDATE embeddings SET vector = ? WHERE \"path\" = ? AND chunk = ?', rows.length);\n if (!rewritten) throw new Error('duckdb: writeVectorBatch SQL shape not recognized by rewriteUpdate');\n\n await withTransaction(conn, async () => {\n const stmt = await duckdb.prepare(rewritten.sql);\n try {\n const values: DuckDBValue[] = [];\n const types: DuckDBType[] = [];\n for (const row of rows) {\n values.push(arrayValue(dequantize(row, dims)), row.path, row.chunk);\n types.push(ARRAY(DOUBLE, dims), untyped, untyped);\n }\n stmt.bind(values, types);\n await stmt.run();\n } finally {\n stmt.destroySync();\n }\n });\n}\n\n// Best chunk per file by cosine, with the earliest chunk winning exact score ties.\n// An empty `allowed` set matches sqlite's scan (every row filtered out) without touching the table.\nexport async function scanCandidates(duckdb: DuckDBConnection, qv: Float32Array, dims: number, fetch: number, allowed?: Set<string>): Promise<VectorCandidate[]> {\n if (allowed && allowed.size === 0) return [];\n const { arrayValue, ARRAY, DOUBLE } = await duckdbApi();\n const allowedList = allowed ? [...allowed] : [];\n const query = padded(qv, dims);\n const queryNorm = Math.sqrt(query.reduce((sum, value) => sum + value * value, 0));\n const score = queryNorm === 0 ? '0.0' : 'CASE WHEN array_inner_product(vector, vector) = 0 THEN 0.0 ELSE array_cosine_similarity(vector, ?) END';\n const sql = `SELECT \"path\", start_line, end_line, score\n FROM (\n SELECT \"path\", start_line, end_line, score,\n row_number() OVER (PARTITION BY \"path\" ORDER BY score DESC, chunk ASC) AS selected\n FROM (\n SELECT \"path\", chunk, start_line, end_line, ${score} AS score\n FROM embeddings\n WHERE vector IS NOT NULL ${allowed ? `AND ${inClause('\"path\"', allowedList.length)}` : ''}\n ) sub\n ) ranked\n WHERE selected = 1\n ORDER BY score DESC, \"path\" ASC\n LIMIT ?`;\n const stmt = await duckdb.prepare(sql);\n try {\n const values: DuckDBValue[] = [...(queryNorm === 0 ? [] : [arrayValue(query)]), ...allowedList, fetch];\n const types: DuckDBType[] = [...(queryNorm === 0 ? [] : [ARRAY(DOUBLE, dims)]), ...allowedList.map(() => untyped), untyped];\n stmt.bind(values, types);\n const reader = await stmt.runAndReadAll();\n const rows = reader.getRowObjectsJS() as Array<{ path: string; start_line: number; end_line: number; score: number }>;\n return rows.map((r) => ({ path: r.path, lines: `L${r.start_line}-${r.end_line}`, similarity: asCosine(r.score) }));\n } finally {\n stmt.destroySync();\n }\n}\n\n// Note-to-note similarity is the max cosine over (target chunk, other chunk) pairs, pushed into one native scan: the sampled target vectors become\n// a small VALUES list, cross-joined against every stored vector and reduced with max()/GROUP BY, so the O(target x stored) cost (12.7s at 201 chunks/note in JS) runs vectorized instead of scalar.\nexport async function scanSimilar(duckdb: DuckDBConnection, conn: Connection, dims: number, path: string, opts: { exclude: Set<string>; allowed?: Set<string>; k: number }): Promise<VectorSimilar[]> {\n if (opts.allowed && opts.allowed.size === 0) return [];\n const targetStmt = await conn.prepare('SELECT vector FROM embeddings WHERE \"path\" = ? AND vector IS NOT NULL ORDER BY chunk');\n const targetRows = (await targetStmt.all(path)) as Array<{ vector: number[] }>;\n if (targetRows.length === 0) return [];\n const targets = sampleEvenly(targetRows).map((row) => row.vector);\n\n const { arrayValue, ARRAY, DOUBLE } = await duckdbApi();\n const excludeList = [...opts.exclude];\n const allowedList = opts.allowed ? [...opts.allowed] : [];\n const sql = `WITH targets(tv) AS (VALUES ${targets.map(() => '(?)').join(', ')}), grouped AS (\n SELECT e.\"path\" AS path, max(CASE WHEN array_inner_product(e.vector, e.vector) = 0 OR array_inner_product(t.tv, t.tv) = 0 THEN 0.0 ELSE array_cosine_similarity(e.vector, t.tv) END) AS score\n FROM embeddings e, targets t\n WHERE e.vector IS NOT NULL AND e.\"path\" != ?\n ${excludeList.length > 0 ? `AND NOT ${inClause('e.\"path\"', excludeList.length)}` : ''}\n ${opts.allowed ? `AND ${inClause('e.\"path\"', allowedList.length)}` : ''}\n GROUP BY e.\"path\"\n )\n SELECT path, score FROM grouped\n ORDER BY score DESC, path ASC\n LIMIT ?`;\n const stmt = await duckdb.prepare(sql);\n try {\n const values: DuckDBValue[] = [...targets.map((t) => arrayValue(padded(t, dims))), path, ...excludeList, ...allowedList, opts.k];\n const types: DuckDBType[] = [...targets.map(() => ARRAY(DOUBLE, dims)), untyped, ...excludeList.map(() => untyped), ...allowedList.map(() => untyped), untyped];\n stmt.bind(values, types);\n const reader = await stmt.runAndReadAll();\n const rows = reader.getRowObjectsJS() as Array<{ path: string; score: number }>;\n return rows.map((r) => ({ path: r.path, similarity: asCosine(r.score) }));\n } finally {\n stmt.destroySync();\n }\n}\n"],"names":["withTransaction","asCosine","sampleEvenly","rewriteUpdate","duckdbApi","untyped","undefined","padded","values","dims","out","Array","fill","d","Math","min","length","inClause","column","count","from","join","dequantize","row","q","Int8Array","vector","buffer","byteOffset","byteLength","v","scale","writeVectorBatch","duckdb","conn","rows","arrayValue","ARRAY","DOUBLE","rewritten","Error","stmt","prepare","sql","types","push","path","chunk","bind","run","destroySync","scanCandidates","qv","fetch","allowed","size","allowedList","query","queryNorm","sqrt","reduce","sum","value","score","map","reader","runAndReadAll","getRowObjectsJS","r","lines","start_line","end_line","similarity","scanSimilar","opts","targetStmt","targetRows","all","targets","excludeList","exclude","t","k"],"mappings":"AACA,SAASA,eAAe,QAAQ,oBAAoB;AAEpD,SAASC,QAAQ,EAAEC,YAAY,QAAQ,gBAAgB;AACvD,SAASC,aAAa,QAAQ,aAAa;AAC3C,SAASC,SAAS,QAAQ,cAAc;AAExC,iJAAiJ;AACjJ,+RAA+R;AAE/R,wIAAwI;AACxI,iIAAiI;AAEjI,6FAA6F;AAC7F,6FAA6F;AAC7F,MAAMC,UAAUC;AAEhB,wHAAwH;AACxH,oIAAoI;AACpI,SAASC,OAAOC,MAAyB,EAAEC,IAAY;IACrD,MAAMC,MAAM,IAAIC,MAAcF,MAAMG,IAAI,CAAC;IACzC,IAAK,IAAIC,IAAI,GAAGA,IAAIC,KAAKC,GAAG,CAACP,OAAOQ,MAAM,EAAEP,OAAOI,IAAKH,GAAG,CAACG,EAAE,GAAGL,MAAM,CAACK,EAAE;IAC1E,OAAOH;AACT;AAEA,SAASO,SAASC,MAAc,EAAEC,KAAa;IAC7C,OAAO,GAAGD,OAAO,KAAK,EAAEP,MAAMS,IAAI,CAAC;QAAEJ,QAAQG;IAAM,GAAG,IAAM,KAAKE,IAAI,CAAC,MAAM,CAAC,CAAC;AAChF;AAEA,6HAA6H;AAC7H,qJAAqJ;AACrJ,SAASC,WAAWC,GAAmB,EAAEd,IAAY;IACnD,MAAMe,IAAI,IAAIC,UAAUF,IAAIG,MAAM,CAACC,MAAM,EAAEJ,IAAIG,MAAM,CAACE,UAAU,EAAEL,IAAIG,MAAM,CAACG,UAAU;IACvF,OAAOtB,OACLI,MAAMS,IAAI,CAACI,GAAG,CAACM,IAAMA,IAAIP,IAAIQ,KAAK,GAClCtB;AAEJ;AAEA,oIAAoI;AACpI,8QAA8Q;AAC9Q,OAAO,eAAeuB,iBAAiBC,MAAwB,EAAEC,IAAgB,EAAEzB,IAAY,EAAE0B,IAAsB;IACrH,IAAIA,KAAKnB,MAAM,KAAK,GAAG;IACvB,MAAM,EAAEoB,UAAU,EAAEC,KAAK,EAAEC,MAAM,EAAE,GAAG,MAAMlC;IAC5C,MAAMmC,YAAYpC,cAAc,mEAAmEgC,KAAKnB,MAAM;IAC9G,IAAI,CAACuB,WAAW,MAAM,IAAIC,MAAM;IAEhC,MAAMxC,gBAAgBkC,MAAM;QAC1B,MAAMO,OAAO,MAAMR,OAAOS,OAAO,CAACH,UAAUI,GAAG;QAC/C,IAAI;YACF,MAAMnC,SAAwB,EAAE;YAChC,MAAMoC,QAAsB,EAAE;YAC9B,KAAK,MAAMrB,OAAOY,KAAM;gBACtB3B,OAAOqC,IAAI,CAACT,WAAWd,WAAWC,KAAKd,QAAQc,IAAIuB,IAAI,EAAEvB,IAAIwB,KAAK;gBAClEH,MAAMC,IAAI,CAACR,MAAMC,QAAQ7B,OAAOJ,SAASA;YAC3C;YACAoC,KAAKO,IAAI,CAACxC,QAAQoC;YAClB,MAAMH,KAAKQ,GAAG;QAChB,SAAU;YACRR,KAAKS,WAAW;QAClB;IACF;AACF;AAEA,mFAAmF;AACnF,oGAAoG;AACpG,OAAO,eAAeC,eAAelB,MAAwB,EAAEmB,EAAgB,EAAE3C,IAAY,EAAE4C,KAAa,EAAEC,OAAqB;IACjI,IAAIA,WAAWA,QAAQC,IAAI,KAAK,GAAG,OAAO,EAAE;IAC5C,MAAM,EAAEnB,UAAU,EAAEC,KAAK,EAAEC,MAAM,EAAE,GAAG,MAAMlC;IAC5C,MAAMoD,cAAcF,UAAU;WAAIA;KAAQ,GAAG,EAAE;IAC/C,MAAMG,QAAQlD,OAAO6C,IAAI3C;IACzB,MAAMiD,YAAY5C,KAAK6C,IAAI,CAACF,MAAMG,MAAM,CAAC,CAACC,KAAKC,QAAUD,MAAMC,QAAQA,OAAO;IAC9E,MAAMC,QAAQL,cAAc,IAAI,QAAQ;IACxC,MAAMf,MAAM,CAAC;;;;;oDAKqC,EAAEoB,MAAM;;iCAE3B,EAAET,UAAU,CAAC,IAAI,EAAErC,SAAS,UAAUuC,YAAYxC,MAAM,GAAG,GAAG,GAAG;;;;;WAKvF,CAAC;IACV,MAAMyB,OAAO,MAAMR,OAAOS,OAAO,CAACC;IAClC,IAAI;QACF,MAAMnC,SAAwB;eAAKkD,cAAc,IAAI,EAAE,GAAG;gBAACtB,WAAWqB;aAAO;eAAMD;YAAaH;SAAM;QACtG,MAAMT,QAAsB;eAAKc,cAAc,IAAI,EAAE,GAAG;gBAACrB,MAAMC,QAAQ7B;aAAM;eAAM+C,YAAYQ,GAAG,CAAC,IAAM3D;YAAUA;SAAQ;QAC3HoC,KAAKO,IAAI,CAACxC,QAAQoC;QAClB,MAAMqB,SAAS,MAAMxB,KAAKyB,aAAa;QACvC,MAAM/B,OAAO8B,OAAOE,eAAe;QACnC,OAAOhC,KAAK6B,GAAG,CAAC,CAACI,IAAO,CAAA;gBAAEtB,MAAMsB,EAAEtB,IAAI;gBAAEuB,OAAO,CAAC,CAAC,EAAED,EAAEE,UAAU,CAAC,CAAC,EAAEF,EAAEG,QAAQ,EAAE;gBAAEC,YAAYvE,SAASmE,EAAEL,KAAK;YAAE,CAAA;IACjH,SAAU;QACRtB,KAAKS,WAAW;IAClB;AACF;AAEA,mJAAmJ;AACnJ,oMAAoM;AACpM,OAAO,eAAeuB,YAAYxC,MAAwB,EAAEC,IAAgB,EAAEzB,IAAY,EAAEqC,IAAY,EAAE4B,IAAgE;IACxK,IAAIA,KAAKpB,OAAO,IAAIoB,KAAKpB,OAAO,CAACC,IAAI,KAAK,GAAG,OAAO,EAAE;IACtD,MAAMoB,aAAa,MAAMzC,KAAKQ,OAAO,CAAC;IACtC,MAAMkC,aAAc,MAAMD,WAAWE,GAAG,CAAC/B;IACzC,IAAI8B,WAAW5D,MAAM,KAAK,GAAG,OAAO,EAAE;IACtC,MAAM8D,UAAU5E,aAAa0E,YAAYZ,GAAG,CAAC,CAACzC,MAAQA,IAAIG,MAAM;IAEhE,MAAM,EAAEU,UAAU,EAAEC,KAAK,EAAEC,MAAM,EAAE,GAAG,MAAMlC;IAC5C,MAAM2E,cAAc;WAAIL,KAAKM,OAAO;KAAC;IACrC,MAAMxB,cAAckB,KAAKpB,OAAO,GAAG;WAAIoB,KAAKpB,OAAO;KAAC,GAAG,EAAE;IACzD,MAAMX,MAAM,CAAC,4BAA4B,EAAEmC,QAAQd,GAAG,CAAC,IAAM,OAAO3C,IAAI,CAAC,MAAM;;;;MAI3E,EAAE0D,YAAY/D,MAAM,GAAG,IAAI,CAAC,QAAQ,EAAEC,SAAS,YAAY8D,YAAY/D,MAAM,GAAG,GAAG,GAAG;MACtF,EAAE0D,KAAKpB,OAAO,GAAG,CAAC,IAAI,EAAErC,SAAS,YAAYuC,YAAYxC,MAAM,GAAG,GAAG,GAAG;;;;;WAKnE,CAAC;IACV,MAAMyB,OAAO,MAAMR,OAAOS,OAAO,CAACC;IAClC,IAAI;QACF,MAAMnC,SAAwB;eAAIsE,QAAQd,GAAG,CAAC,CAACiB,IAAM7C,WAAW7B,OAAO0E,GAAGxE;YAASqC;eAASiC;eAAgBvB;YAAakB,KAAKQ,CAAC;SAAC;QAChI,MAAMtC,QAAsB;eAAIkC,QAAQd,GAAG,CAAC,IAAM3B,MAAMC,QAAQ7B;YAAQJ;eAAY0E,YAAYf,GAAG,CAAC,IAAM3D;eAAamD,YAAYQ,GAAG,CAAC,IAAM3D;YAAUA;SAAQ;QAC/JoC,KAAKO,IAAI,CAACxC,QAAQoC;QAClB,MAAMqB,SAAS,MAAMxB,KAAKyB,aAAa;QACvC,MAAM/B,OAAO8B,OAAOE,eAAe;QACnC,OAAOhC,KAAK6B,GAAG,CAAC,CAACI,IAAO,CAAA;gBAAEtB,MAAMsB,EAAEtB,IAAI;gBAAE0B,YAAYvE,SAASmE,EAAEL,KAAK;YAAE,CAAA;IACxE,SAAU;QACRtB,KAAKS,WAAW;IAClB;AACF"}
|
|
1
|
+
{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/store/duckdb/vectors.ts"],"sourcesContent":["import type { DuckDBConnection, DuckDBType, DuckDBValue } from '@duckdb/node-api';\nimport { withTransaction } from '../transaction.ts';\nimport type { Connection, VectorCandidate, VectorSimilar, VectorWriteRow } from '../types.ts';\nimport { asCosine, sampleEvenly } from '../vectors.ts';\nimport { duckdbApi } from './native.ts';\n\n// This store keeps vectors as native FLOAT[dims] arrays (dims fixed at DDL time by open.ts's ensureSchema, from embed/types.ts's STORE_DIMS) and\n// scans them with array_cosine_similarity, pushing top-k into SQL instead of pulling every row into JS; sqlite's int8+scale BLOB scan is its own representation. Every function here takes `dims` as a parameter rather than importing STORE_DIMS directly, so it stays testable at any width.\n\n// A value import of '@duckdb/node-api' must never sit at module top level, or a sqlite-only tree would resolve this optional dependency\n// just by importing store/index.ts; native.ts's shared duckdbApi() keeps that lazy and reuses whatever open.ts already resolved.\n\n// A bind position whose type is left to auto-inference (safe for plain strings/numbers; only\n// the vector ARRAY positions below need an explicit type -- see writeVectorBatch's comment).\nconst untyped = undefined as unknown as DuckDBType;\nconst VECTOR_WRITE_STAGE = '_sense_vector_write_stage';\n\n// Created once per native connection outside write transactions; its lifetime follows that connection.\nexport async function createVectorWriteStage(duckdb: DuckDBConnection, dims: number): Promise<void> {\n await duckdb.run(`CREATE TEMP TABLE ${VECTOR_WRITE_STAGE} (\"path\" TEXT, chunk INTEGER, vector FLOAT[${dims}])`);\n}\n\n// The DDL-fixed array width can exceed a vector's actual length (a hypothetical model sliced under the column's width);\n// zero-padding leaves cosine scores unchanged since the added dimensions contribute nothing to either vector's dot product or norm.\nfunction padded(values: ArrayLike<number>, dims: number): number[] {\n const out = new Array<number>(dims).fill(0);\n for (let d = 0; d < Math.min(values.length, dims); d++) out[d] = values[d];\n return out;\n}\n\nfunction inClause(column: string, count: number): string {\n return `${column} IN (${Array.from({ length: count }, () => '?').join(', ')})`;\n}\n\n// Rows arrive int8-quantized with a per-vector scale (the wire format embed/query.ts produces for every store, see toStore);\n// dequantizing into a plain float array is this store's own representation choice, so both stores are handed the same vectors and diverge only here.\nfunction dequantize(row: VectorWriteRow, dims: number): number[] {\n const q = new Int8Array(row.vector.buffer, row.vector.byteOffset, row.vector.byteLength);\n return padded(\n Array.from(q, (v) => v * row.scale),\n dims\n );\n}\n\n// The native appender needs an explicit FLOAT array type: inference from an integer first component\n// silently truncates fractional components. The fixed UPDATE keeps each provider batch keyed by path and chunk.\nexport async function writeVectorBatch(duckdb: DuckDBConnection, conn: Connection, dims: number, rows: VectorWriteRow[]): Promise<void> {\n if (rows.length === 0) return;\n const { ARRAY, FLOAT } = await duckdbApi();\n const vectorType = ARRAY(FLOAT, dims);\n\n await withTransaction(conn, async () => {\n await duckdb.run(`DELETE FROM ${VECTOR_WRITE_STAGE}`);\n const appender = await duckdb.createAppender(VECTOR_WRITE_STAGE);\n try {\n for (const row of rows) {\n appender.appendVarchar(row.path);\n appender.appendInteger(row.chunk);\n appender.appendArray(dequantize(row, dims), vectorType);\n appender.endRow();\n }\n appender.flushSync();\n } finally {\n appender.closeSync();\n }\n await duckdb.run(`UPDATE embeddings SET vector = staged.vector FROM ${VECTOR_WRITE_STAGE} AS staged WHERE embeddings.\"path\" = staged.\"path\" AND embeddings.chunk = staged.chunk`);\n });\n}\n\n// Best chunk per file by cosine, with the earliest chunk winning exact score ties.\n// An empty `allowed` set matches sqlite's scan (every row filtered out) without touching the table.\nexport async function scanCandidates(duckdb: DuckDBConnection, qv: Float32Array, dims: number, fetch: number, allowed?: Set<string>): Promise<VectorCandidate[]> {\n if (allowed && allowed.size === 0) return [];\n const { arrayValue, ARRAY, DOUBLE } = await duckdbApi();\n const allowedList = allowed ? [...allowed] : [];\n const query = padded(qv, dims);\n const queryNorm = Math.sqrt(query.reduce((sum, value) => sum + value * value, 0));\n const score = queryNorm === 0 ? '0.0' : 'CASE WHEN array_inner_product(vector, vector) = 0 THEN 0.0 ELSE array_cosine_similarity(vector, ?) END';\n const sql = `SELECT \"path\", start_line, end_line, score\n FROM (\n SELECT \"path\", start_line, end_line, score,\n row_number() OVER (PARTITION BY \"path\" ORDER BY score DESC, chunk ASC) AS selected\n FROM (\n SELECT \"path\", chunk, start_line, end_line, ${score} AS score\n FROM embeddings\n WHERE vector IS NOT NULL ${allowed ? `AND ${inClause('\"path\"', allowedList.length)}` : ''}\n ) sub\n ) ranked\n WHERE selected = 1\n ORDER BY score DESC, \"path\" ASC\n LIMIT ?`;\n const stmt = await duckdb.prepare(sql);\n try {\n const values: DuckDBValue[] = [...(queryNorm === 0 ? [] : [arrayValue(query)]), ...allowedList, fetch];\n const types: DuckDBType[] = [...(queryNorm === 0 ? [] : [ARRAY(DOUBLE, dims)]), ...allowedList.map(() => untyped), untyped];\n stmt.bind(values, types);\n const reader = await stmt.runAndReadAll();\n const rows = reader.getRowObjectsJS() as Array<{ path: string; start_line: number; end_line: number; score: number }>;\n return rows.map((r) => ({ path: r.path, lines: `L${r.start_line}-${r.end_line}`, similarity: asCosine(r.score) }));\n } finally {\n stmt.destroySync();\n }\n}\n\n// Note-to-note similarity is the max cosine over (target chunk, other chunk) pairs, pushed into one native scan: the sampled target vectors become\n// a small VALUES list, cross-joined against every stored vector and reduced with max()/GROUP BY, so the O(target x stored) cost (12.7s at 201 chunks/note in JS) runs vectorized instead of scalar.\nexport async function scanSimilar(duckdb: DuckDBConnection, conn: Connection, dims: number, path: string, opts: { exclude: Set<string>; allowed?: Set<string>; k: number }): Promise<VectorSimilar[]> {\n if (opts.allowed && opts.allowed.size === 0) return [];\n const targetStmt = await conn.prepare('SELECT vector FROM embeddings WHERE \"path\" = ? AND vector IS NOT NULL ORDER BY chunk');\n const targetRows = (await targetStmt.all(path)) as Array<{ vector: number[] }>;\n if (targetRows.length === 0) return [];\n const targets = sampleEvenly(targetRows).map((row) => row.vector);\n\n const { arrayValue, ARRAY, DOUBLE } = await duckdbApi();\n const excludeList = [...opts.exclude];\n const allowedList = opts.allowed ? [...opts.allowed] : [];\n const sql = `WITH targets(tv) AS (VALUES ${targets.map(() => '(?)').join(', ')}), grouped AS (\n SELECT e.\"path\" AS path, max(CASE WHEN array_inner_product(e.vector, e.vector) = 0 OR array_inner_product(t.tv, t.tv) = 0 THEN 0.0 ELSE array_cosine_similarity(e.vector, t.tv) END) AS score\n FROM embeddings e, targets t\n WHERE e.vector IS NOT NULL AND e.\"path\" != ?\n ${excludeList.length > 0 ? `AND NOT ${inClause('e.\"path\"', excludeList.length)}` : ''}\n ${opts.allowed ? `AND ${inClause('e.\"path\"', allowedList.length)}` : ''}\n GROUP BY e.\"path\"\n )\n SELECT path, score FROM grouped\n ORDER BY score DESC, path ASC\n LIMIT ?`;\n const stmt = await duckdb.prepare(sql);\n try {\n const values: DuckDBValue[] = [...targets.map((t) => arrayValue(padded(t, dims))), path, ...excludeList, ...allowedList, opts.k];\n const types: DuckDBType[] = [...targets.map(() => ARRAY(DOUBLE, dims)), untyped, ...excludeList.map(() => untyped), ...allowedList.map(() => untyped), untyped];\n stmt.bind(values, types);\n const reader = await stmt.runAndReadAll();\n const rows = reader.getRowObjectsJS() as Array<{ path: string; score: number }>;\n return rows.map((r) => ({ path: r.path, similarity: asCosine(r.score) }));\n } finally {\n stmt.destroySync();\n }\n}\n"],"names":["withTransaction","asCosine","sampleEvenly","duckdbApi","untyped","undefined","VECTOR_WRITE_STAGE","createVectorWriteStage","duckdb","dims","run","padded","values","out","Array","fill","d","Math","min","length","inClause","column","count","from","join","dequantize","row","q","Int8Array","vector","buffer","byteOffset","byteLength","v","scale","writeVectorBatch","conn","rows","ARRAY","FLOAT","vectorType","appender","createAppender","appendVarchar","path","appendInteger","chunk","appendArray","endRow","flushSync","closeSync","scanCandidates","qv","fetch","allowed","size","arrayValue","DOUBLE","allowedList","query","queryNorm","sqrt","reduce","sum","value","score","sql","stmt","prepare","types","map","bind","reader","runAndReadAll","getRowObjectsJS","r","lines","start_line","end_line","similarity","destroySync","scanSimilar","opts","targetStmt","targetRows","all","targets","excludeList","exclude","t","k"],"mappings":"AACA,SAASA,eAAe,QAAQ,oBAAoB;AAEpD,SAASC,QAAQ,EAAEC,YAAY,QAAQ,gBAAgB;AACvD,SAASC,SAAS,QAAQ,cAAc;AAExC,iJAAiJ;AACjJ,+RAA+R;AAE/R,wIAAwI;AACxI,iIAAiI;AAEjI,6FAA6F;AAC7F,6FAA6F;AAC7F,MAAMC,UAAUC;AAChB,MAAMC,qBAAqB;AAE3B,uGAAuG;AACvG,OAAO,eAAeC,uBAAuBC,MAAwB,EAAEC,IAAY;IACjF,MAAMD,OAAOE,GAAG,CAAC,CAAC,kBAAkB,EAAEJ,mBAAmB,2CAA2C,EAAEG,KAAK,EAAE,CAAC;AAChH;AAEA,wHAAwH;AACxH,oIAAoI;AACpI,SAASE,OAAOC,MAAyB,EAAEH,IAAY;IACrD,MAAMI,MAAM,IAAIC,MAAcL,MAAMM,IAAI,CAAC;IACzC,IAAK,IAAIC,IAAI,GAAGA,IAAIC,KAAKC,GAAG,CAACN,OAAOO,MAAM,EAAEV,OAAOO,IAAKH,GAAG,CAACG,EAAE,GAAGJ,MAAM,CAACI,EAAE;IAC1E,OAAOH;AACT;AAEA,SAASO,SAASC,MAAc,EAAEC,KAAa;IAC7C,OAAO,GAAGD,OAAO,KAAK,EAAEP,MAAMS,IAAI,CAAC;QAAEJ,QAAQG;IAAM,GAAG,IAAM,KAAKE,IAAI,CAAC,MAAM,CAAC,CAAC;AAChF;AAEA,6HAA6H;AAC7H,qJAAqJ;AACrJ,SAASC,WAAWC,GAAmB,EAAEjB,IAAY;IACnD,MAAMkB,IAAI,IAAIC,UAAUF,IAAIG,MAAM,CAACC,MAAM,EAAEJ,IAAIG,MAAM,CAACE,UAAU,EAAEL,IAAIG,MAAM,CAACG,UAAU;IACvF,OAAOrB,OACLG,MAAMS,IAAI,CAACI,GAAG,CAACM,IAAMA,IAAIP,IAAIQ,KAAK,GAClCzB;AAEJ;AAEA,oGAAoG;AACpG,gHAAgH;AAChH,OAAO,eAAe0B,iBAAiB3B,MAAwB,EAAE4B,IAAgB,EAAE3B,IAAY,EAAE4B,IAAsB;IACrH,IAAIA,KAAKlB,MAAM,KAAK,GAAG;IACvB,MAAM,EAAEmB,KAAK,EAAEC,KAAK,EAAE,GAAG,MAAMpC;IAC/B,MAAMqC,aAAaF,MAAMC,OAAO9B;IAEhC,MAAMT,gBAAgBoC,MAAM;QAC1B,MAAM5B,OAAOE,GAAG,CAAC,CAAC,YAAY,EAAEJ,oBAAoB;QACpD,MAAMmC,WAAW,MAAMjC,OAAOkC,cAAc,CAACpC;QAC7C,IAAI;YACF,KAAK,MAAMoB,OAAOW,KAAM;gBACtBI,SAASE,aAAa,CAACjB,IAAIkB,IAAI;gBAC/BH,SAASI,aAAa,CAACnB,IAAIoB,KAAK;gBAChCL,SAASM,WAAW,CAACtB,WAAWC,KAAKjB,OAAO+B;gBAC5CC,SAASO,MAAM;YACjB;YACAP,SAASQ,SAAS;QACpB,SAAU;YACRR,SAASS,SAAS;QACpB;QACA,MAAM1C,OAAOE,GAAG,CAAC,CAAC,kDAAkD,EAAEJ,mBAAmB,sFAAsF,CAAC;IAClL;AACF;AAEA,mFAAmF;AACnF,oGAAoG;AACpG,OAAO,eAAe6C,eAAe3C,MAAwB,EAAE4C,EAAgB,EAAE3C,IAAY,EAAE4C,KAAa,EAAEC,OAAqB;IACjI,IAAIA,WAAWA,QAAQC,IAAI,KAAK,GAAG,OAAO,EAAE;IAC5C,MAAM,EAAEC,UAAU,EAAElB,KAAK,EAAEmB,MAAM,EAAE,GAAG,MAAMtD;IAC5C,MAAMuD,cAAcJ,UAAU;WAAIA;KAAQ,GAAG,EAAE;IAC/C,MAAMK,QAAQhD,OAAOyC,IAAI3C;IACzB,MAAMmD,YAAY3C,KAAK4C,IAAI,CAACF,MAAMG,MAAM,CAAC,CAACC,KAAKC,QAAUD,MAAMC,QAAQA,OAAO;IAC9E,MAAMC,QAAQL,cAAc,IAAI,QAAQ;IACxC,MAAMM,MAAM,CAAC;;;;;oDAKqC,EAAED,MAAM;;iCAE3B,EAAEX,UAAU,CAAC,IAAI,EAAElC,SAAS,UAAUsC,YAAYvC,MAAM,GAAG,GAAG,GAAG;;;;;WAKvF,CAAC;IACV,MAAMgD,OAAO,MAAM3D,OAAO4D,OAAO,CAACF;IAClC,IAAI;QACF,MAAMtD,SAAwB;eAAKgD,cAAc,IAAI,EAAE,GAAG;gBAACJ,WAAWG;aAAO;eAAMD;YAAaL;SAAM;QACtG,MAAMgB,QAAsB;eAAKT,cAAc,IAAI,EAAE,GAAG;gBAACtB,MAAMmB,QAAQhD;aAAM;eAAMiD,YAAYY,GAAG,CAAC,IAAMlE;YAAUA;SAAQ;QAC3H+D,KAAKI,IAAI,CAAC3D,QAAQyD;QAClB,MAAMG,SAAS,MAAML,KAAKM,aAAa;QACvC,MAAMpC,OAAOmC,OAAOE,eAAe;QACnC,OAAOrC,KAAKiC,GAAG,CAAC,CAACK,IAAO,CAAA;gBAAE/B,MAAM+B,EAAE/B,IAAI;gBAAEgC,OAAO,CAAC,CAAC,EAAED,EAAEE,UAAU,CAAC,CAAC,EAAEF,EAAEG,QAAQ,EAAE;gBAAEC,YAAY9E,SAAS0E,EAAEV,KAAK;YAAE,CAAA;IACjH,SAAU;QACRE,KAAKa,WAAW;IAClB;AACF;AAEA,mJAAmJ;AACnJ,oMAAoM;AACpM,OAAO,eAAeC,YAAYzE,MAAwB,EAAE4B,IAAgB,EAAE3B,IAAY,EAAEmC,IAAY,EAAEsC,IAAgE;IACxK,IAAIA,KAAK5B,OAAO,IAAI4B,KAAK5B,OAAO,CAACC,IAAI,KAAK,GAAG,OAAO,EAAE;IACtD,MAAM4B,aAAa,MAAM/C,KAAKgC,OAAO,CAAC;IACtC,MAAMgB,aAAc,MAAMD,WAAWE,GAAG,CAACzC;IACzC,IAAIwC,WAAWjE,MAAM,KAAK,GAAG,OAAO,EAAE;IACtC,MAAMmE,UAAUpF,aAAakF,YAAYd,GAAG,CAAC,CAAC5C,MAAQA,IAAIG,MAAM;IAEhE,MAAM,EAAE2B,UAAU,EAAElB,KAAK,EAAEmB,MAAM,EAAE,GAAG,MAAMtD;IAC5C,MAAMoF,cAAc;WAAIL,KAAKM,OAAO;KAAC;IACrC,MAAM9B,cAAcwB,KAAK5B,OAAO,GAAG;WAAI4B,KAAK5B,OAAO;KAAC,GAAG,EAAE;IACzD,MAAMY,MAAM,CAAC,4BAA4B,EAAEoB,QAAQhB,GAAG,CAAC,IAAM,OAAO9C,IAAI,CAAC,MAAM;;;;MAI3E,EAAE+D,YAAYpE,MAAM,GAAG,IAAI,CAAC,QAAQ,EAAEC,SAAS,YAAYmE,YAAYpE,MAAM,GAAG,GAAG,GAAG;MACtF,EAAE+D,KAAK5B,OAAO,GAAG,CAAC,IAAI,EAAElC,SAAS,YAAYsC,YAAYvC,MAAM,GAAG,GAAG,GAAG;;;;;WAKnE,CAAC;IACV,MAAMgD,OAAO,MAAM3D,OAAO4D,OAAO,CAACF;IAClC,IAAI;QACF,MAAMtD,SAAwB;eAAI0E,QAAQhB,GAAG,CAAC,CAACmB,IAAMjC,WAAW7C,OAAO8E,GAAGhF;YAASmC;eAAS2C;eAAgB7B;YAAawB,KAAKQ,CAAC;SAAC;QAChI,MAAMrB,QAAsB;eAAIiB,QAAQhB,GAAG,CAAC,IAAMhC,MAAMmB,QAAQhD;YAAQL;eAAYmF,YAAYjB,GAAG,CAAC,IAAMlE;eAAasD,YAAYY,GAAG,CAAC,IAAMlE;YAAUA;SAAQ;QAC/J+D,KAAKI,IAAI,CAAC3D,QAAQyD;QAClB,MAAMG,SAAS,MAAML,KAAKM,aAAa;QACvC,MAAMpC,OAAOmC,OAAOE,eAAe;QACnC,OAAOrC,KAAKiC,GAAG,CAAC,CAACK,IAAO,CAAA;gBAAE/B,MAAM+B,EAAE/B,IAAI;gBAAEmC,YAAY9E,SAAS0E,EAAEV,KAAK;YAAE,CAAA;IACxE,SAAU;QACRE,KAAKa,WAAW;IAClB;AACF"}
|
package/dist/esm/watch.js
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
import { watch as fsWatch } from 'node:fs';
|
|
2
|
-
import { STATE_DIR } from './config/index.js';
|
|
3
2
|
import { SenseError } from './errors.js';
|
|
4
3
|
import { guardedTick } from './lib/guarded-tick.js';
|
|
4
|
+
import { shouldReconcileWatchNotification } from './lib/watch-notification.js';
|
|
5
5
|
import { docCount, openStore } from './store/index.js';
|
|
6
6
|
import { startWatchClaim, WATCH_HEARTBEAT_INTERVAL_MS } from './watch-claim.js';
|
|
7
7
|
// Watch is a cache pre-warmer, not a correctness mechanism: open() always reconciles anyway, so any fs event just triggers a debounced full reconcile.
|
|
@@ -97,7 +97,7 @@ export async function runWatch(cfg, opts = {}) {
|
|
|
97
97
|
const watcher = fsWatch(rootDir, {
|
|
98
98
|
recursive: true
|
|
99
99
|
}, (_event, filename)=>{
|
|
100
|
-
if (
|
|
100
|
+
if (!shouldReconcileWatchNotification(filename)) return;
|
|
101
101
|
scheduleReconcile();
|
|
102
102
|
});
|
|
103
103
|
const reconcileTimer = setInterval(guardedTick(async ()=>requestCycle(false), ()=>stopping), reconcileIntervalMs);
|
package/dist/esm/watch.js.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/watch.ts"],"sourcesContent":["import { watch as fsWatch } from 'node:fs';\nimport type { ResolvedConfig } from './config/index.ts';\nimport { STATE_DIR } from './config/index.ts';\nimport { SenseError } from './errors.ts';\nimport { guardedTick } from './lib/guarded-tick.ts';\nimport { docCount, openStore } from './store/index.ts';\nimport { startWatchClaim, WATCH_HEARTBEAT_INTERVAL_MS } from './watch-claim.ts';\n\n// Watch is a cache pre-warmer, not a correctness mechanism: open() always reconciles anyway, so any fs event just triggers a debounced full reconcile.\nconst DEBOUNCE_MS = 200;\n\nexport type WatchEvent = { type: 'started'; rootDir: string; dbPath: string } | { type: 'reconciled'; parsed: number; total: number; warnings: string[] } | { type: 'reconcile-error'; message: string };\n\nexport interface WatchOptions {\n force?: boolean;\n onEvent?: (event: WatchEvent) => void;\n // Aborting runs the same shutdown path as SIGINT/SIGTERM.\n signal?: AbortSignal;\n debounceMs?: number;\n heartbeatIntervalMs?: number;\n}\n\n// Runs in the foreground until SIGINT/SIGTERM/signal abort. DuckDB and Turso hand the cache file to\n// one process at a time, so every reconcile cycle opens and closes instead of holding an idle lock.\nexport async function runWatch(cfg: ResolvedConfig, opts: WatchOptions = {}): Promise<void> {\n const onEvent = opts.onEvent ?? (() => {});\n const debounceMs = opts.debounceMs ?? DEBOUNCE_MS;\n const reconcileIntervalMs = opts.heartbeatIntervalMs ?? WATCH_HEARTBEAT_INTERVAL_MS;\n if (!Number.isFinite(reconcileIntervalMs) || reconcileIntervalMs <= 0) throw new SenseError('CONFIG_INVALID', 'watch heartbeatIntervalMs must be a finite positive number');\n // Reconciliation may be parked for a caller, but live ownership must renew before fixed expiry.\n const claimIntervalMs = Math.min(reconcileIntervalMs, WATCH_HEARTBEAT_INTERVAL_MS);\n const rootDir = cfg.rootDir ?? cfg.baseDir;\n const configDir = cfg.configDir ?? rootDir;\n let claimFailure: Error | null = null;\n let requestShutdown: ((reason: Error) => void) | undefined;\n const claim = await startWatchClaim(configDir, opts.force ?? false, claimIntervalMs, (err) => {\n claimFailure ??= err;\n requestShutdown?.(err);\n });\n let claimCloseStarted = false;\n const closeClaim = (): Promise<void> => {\n claimCloseStarted = true;\n return claim.close();\n };\n try {\n const { store: initialStore, dbPath, warnings: initialWarnings, parsed: initialParsed } = await openStore(cfg);\n let initialTotal = 0;\n try {\n // total is read before close: an event fires only once nothing is held open again.\n initialTotal = initialWarnings.length > 0 || initialParsed > 0 ? await docCount(initialStore) : 0;\n } finally {\n await initialStore.close();\n }\n if (claimFailure) throw claimFailure;\n\n let stopping = false;\n // At most one open+reconcile+close cycle runs at a time; a trigger that arrives mid-cycle is\n // coalesced into a single rerun instead of opening a second, overlapping connection.\n let running: Promise<void> | null = null;\n let queuedAlways: boolean | null = null;\n\n const tick = async (alwaysEmit: boolean): Promise<void> => {\n try {\n const { store, parsed, warnings } = await openStore(cfg);\n try {\n if (alwaysEmit || parsed > 0 || warnings.length > 0) {\n onEvent({ type: 'reconciled', parsed, total: await docCount(store), warnings });\n }\n } finally {\n await store.close();\n }\n } catch (err) {\n onEvent({ type: 'reconcile-error', message: (err as Error).message });\n }\n };\n\n const startCycle = (alwaysEmit: boolean) => {\n running = tick(alwaysEmit).then(() => {\n running = null;\n if (queuedAlways !== null && !stopping) {\n const again = queuedAlways;\n queuedAlways = null;\n startCycle(again);\n }\n });\n };\n\n const requestCycle = (alwaysEmit: boolean) => {\n if (stopping) return;\n if (running) {\n queuedAlways = queuedAlways || alwaysEmit;\n return;\n }\n startCycle(alwaysEmit);\n };\n\n let debounceTimer: NodeJS.Timeout | null = null;\n const scheduleReconcile = () => {\n if (debounceTimer) clearTimeout(debounceTimer);\n debounceTimer = setTimeout(() => {\n debounceTimer = null;\n requestCycle(true);\n }, debounceMs);\n };\n\n // Ignore our own state files, or cache and claim writes would retrigger reconciliation forever.\n // An unresolvable filename reconciles because parsing nothing costs less than missing a real edit.\n const watcher = fsWatch(rootDir, { recursive: true }, (_event, filename) => {\n if (typeof filename === 'string' && filename.startsWith(STATE_DIR)) return;\n scheduleReconcile();\n });\n const reconcileTimer = setInterval(\n guardedTick(\n async () => requestCycle(false),\n () => stopping\n ),\n reconcileIntervalMs\n );\n\n return await new Promise<void>((resolveShutdown, rejectShutdown) => {\n let shutdownReason: Error | null = null;\n const shutdown = (reason?: Error) => {\n shutdownReason ??= reason ?? null;\n if (stopping) return;\n stopping = true;\n const shutdownTask = (async () => {\n process.off('SIGINT', stopNormally);\n process.off('SIGTERM', stopNormally);\n opts.signal?.removeEventListener('abort', stopNormally);\n clearInterval(reconcileTimer);\n if (debounceTimer) clearTimeout(debounceTimer);\n watcher.close();\n await running;\n let finalError: unknown;\n try {\n // A normal stop performs one final freshness pass. Ownership loss drains existing work\n // but starts nothing new because the replacement watcher now owns that responsibility.\n if (!shutdownReason) {\n const { store: finalStore } = await openStore(cfg);\n await finalStore.close();\n }\n } catch (err) {\n finalError = err;\n }\n try {\n await closeClaim();\n } catch (err) {\n if (finalError) throw new AggregateError([finalError, err], 'watch shutdown and claim cleanup both failed');\n throw err;\n }\n if (finalError) throw finalError;\n if (shutdownReason) throw shutdownReason;\n })();\n shutdownTask.then(resolveShutdown, rejectShutdown);\n };\n const stopNormally = () => shutdown();\n requestShutdown = shutdown;\n process.once('SIGINT', stopNormally);\n process.once('SIGTERM', stopNormally);\n if (claimFailure) shutdown(claimFailure);\n else if (opts.signal?.aborted) shutdown();\n else {\n opts.signal?.addEventListener('abort', stopNormally, { once: true });\n try {\n onEvent({ type: 'started', rootDir, dbPath });\n if (!stopping && (initialWarnings.length > 0 || initialParsed > 0)) {\n onEvent({ type: 'reconciled', parsed: initialParsed, total: initialTotal, warnings: initialWarnings });\n }\n } catch (err) {\n shutdown(err instanceof Error ? err : new Error(String(err)));\n }\n }\n });\n } catch (err) {\n if (claimCloseStarted) throw err;\n try {\n await closeClaim();\n } catch (closeError) {\n throw new AggregateError([err, closeError], 'watch setup and claim cleanup both failed');\n }\n throw err;\n } finally {\n requestShutdown = undefined;\n }\n}\n"],"names":["watch","fsWatch","STATE_DIR","SenseError","guardedTick","docCount","openStore","startWatchClaim","WATCH_HEARTBEAT_INTERVAL_MS","DEBOUNCE_MS","runWatch","cfg","opts","onEvent","debounceMs","reconcileIntervalMs","heartbeatIntervalMs","Number","isFinite","claimIntervalMs","Math","min","rootDir","baseDir","configDir","claimFailure","requestShutdown","claim","force","err","claimCloseStarted","closeClaim","close","store","initialStore","dbPath","warnings","initialWarnings","parsed","initialParsed","initialTotal","length","stopping","running","queuedAlways","tick","alwaysEmit","type","total","message","startCycle","then","again","requestCycle","debounceTimer","scheduleReconcile","clearTimeout","setTimeout","watcher","recursive","_event","filename","startsWith","reconcileTimer","setInterval","Promise","resolveShutdown","rejectShutdown","shutdownReason","shutdown","reason","shutdownTask","process","off","stopNormally","signal","removeEventListener","clearInterval","finalError","finalStore","AggregateError","once","aborted","addEventListener","Error","String","closeError","undefined"],"mappings":"AAAA,SAASA,SAASC,OAAO,QAAQ,UAAU;AAE3C,SAASC,SAAS,QAAQ,oBAAoB;AAC9C,SAASC,UAAU,QAAQ,cAAc;AACzC,SAASC,WAAW,QAAQ,wBAAwB;AACpD,SAASC,QAAQ,EAAEC,SAAS,QAAQ,mBAAmB;AACvD,SAASC,eAAe,EAAEC,2BAA2B,QAAQ,mBAAmB;AAEhF,uJAAuJ;AACvJ,MAAMC,cAAc;AAapB,oGAAoG;AACpG,oGAAoG;AACpG,OAAO,eAAeC,SAASC,GAAmB,EAAEC,OAAqB,CAAC,CAAC;QACzDA,eACGA,kBACSA,2BAIZD,cACEA,gBAG6BC;IAV/C,MAAMC,WAAUD,gBAAAA,KAAKC,OAAO,cAAZD,2BAAAA,gBAAiB,KAAO;IACxC,MAAME,cAAaF,mBAAAA,KAAKE,UAAU,cAAfF,8BAAAA,mBAAmBH;IACtC,MAAMM,uBAAsBH,4BAAAA,KAAKI,mBAAmB,cAAxBJ,uCAAAA,4BAA4BJ;IACxD,IAAI,CAACS,OAAOC,QAAQ,CAACH,wBAAwBA,uBAAuB,GAAG,MAAM,IAAIZ,WAAW,kBAAkB;IAC9G,gGAAgG;IAChG,MAAMgB,kBAAkBC,KAAKC,GAAG,CAACN,qBAAqBP;IACtD,MAAMc,WAAUX,eAAAA,IAAIW,OAAO,cAAXX,0BAAAA,eAAeA,IAAIY,OAAO;IAC1C,MAAMC,aAAYb,iBAAAA,IAAIa,SAAS,cAAbb,4BAAAA,iBAAiBW;IACnC,IAAIG,eAA6B;IACjC,IAAIC;IACJ,MAAMC,QAAQ,MAAMpB,gBAAgBiB,YAAWZ,cAAAA,KAAKgB,KAAK,cAAVhB,yBAAAA,cAAc,OAAOO,iBAAiB,CAACU;QACpFJ,yBAAAA,0BAAAA,eAAAA,eAAiBI;QACjBH,4BAAAA,sCAAAA,gBAAkBG;IACpB;IACA,IAAIC,oBAAoB;IACxB,MAAMC,aAAa;QACjBD,oBAAoB;QACpB,OAAOH,MAAMK,KAAK;IACpB;IACA,IAAI;QACF,MAAM,EAAEC,OAAOC,YAAY,EAAEC,MAAM,EAAEC,UAAUC,eAAe,EAAEC,QAAQC,aAAa,EAAE,GAAG,MAAMjC,UAAUK;QAC1G,IAAI6B,eAAe;QACnB,IAAI;YACF,mFAAmF;YACnFA,eAAeH,gBAAgBI,MAAM,GAAG,KAAKF,gBAAgB,IAAI,MAAMlC,SAAS6B,gBAAgB;QAClG,SAAU;YACR,MAAMA,aAAaF,KAAK;QAC1B;QACA,IAAIP,cAAc,MAAMA;QAExB,IAAIiB,WAAW;QACf,6FAA6F;QAC7F,qFAAqF;QACrF,IAAIC,UAAgC;QACpC,IAAIC,eAA+B;QAEnC,MAAMC,OAAO,OAAOC;YAClB,IAAI;gBACF,MAAM,EAAEb,KAAK,EAAEK,MAAM,EAAEF,QAAQ,EAAE,GAAG,MAAM9B,UAAUK;gBACpD,IAAI;oBACF,IAAImC,cAAcR,SAAS,KAAKF,SAASK,MAAM,GAAG,GAAG;wBACnD5B,QAAQ;4BAAEkC,MAAM;4BAAcT;4BAAQU,OAAO,MAAM3C,SAAS4B;4BAAQG;wBAAS;oBAC/E;gBACF,SAAU;oBACR,MAAMH,MAAMD,KAAK;gBACnB;YACF,EAAE,OAAOH,KAAK;gBACZhB,QAAQ;oBAAEkC,MAAM;oBAAmBE,SAAS,AAACpB,IAAcoB,OAAO;gBAAC;YACrE;QACF;QAEA,MAAMC,aAAa,CAACJ;YAClBH,UAAUE,KAAKC,YAAYK,IAAI,CAAC;gBAC9BR,UAAU;gBACV,IAAIC,iBAAiB,QAAQ,CAACF,UAAU;oBACtC,MAAMU,QAAQR;oBACdA,eAAe;oBACfM,WAAWE;gBACb;YACF;QACF;QAEA,MAAMC,eAAe,CAACP;YACpB,IAAIJ,UAAU;YACd,IAAIC,SAAS;gBACXC,eAAeA,gBAAgBE;gBAC/B;YACF;YACAI,WAAWJ;QACb;QAEA,IAAIQ,gBAAuC;QAC3C,MAAMC,oBAAoB;YACxB,IAAID,eAAeE,aAAaF;YAChCA,gBAAgBG,WAAW;gBACzBH,gBAAgB;gBAChBD,aAAa;YACf,GAAGvC;QACL;QAEA,gGAAgG;QAChG,mGAAmG;QACnG,MAAM4C,UAAUzD,QAAQqB,SAAS;YAAEqC,WAAW;QAAK,GAAG,CAACC,QAAQC;YAC7D,IAAI,OAAOA,aAAa,YAAYA,SAASC,UAAU,CAAC5D,YAAY;YACpEqD;QACF;QACA,MAAMQ,iBAAiBC,YACrB5D,YACE,UAAYiD,aAAa,QACzB,IAAMX,WAER3B;QAGF,OAAO,MAAM,IAAIkD,QAAc,CAACC,iBAAiBC;gBAyCtCvD;YAxCT,IAAIwD,iBAA+B;YACnC,MAAMC,WAAW,CAACC;gBAChBF,2BAAAA,4BAAAA,iBAAAA,iBAAmBE,mBAAAA,oBAAAA,SAAU;gBAC7B,IAAI5B,UAAU;gBACdA,WAAW;gBACX,MAAM6B,eAAe,AAAC,CAAA;wBAGpB3D;oBAFA4D,QAAQC,GAAG,CAAC,UAAUC;oBACtBF,QAAQC,GAAG,CAAC,WAAWC;qBACvB9D,eAAAA,KAAK+D,MAAM,cAAX/D,mCAAAA,aAAagE,mBAAmB,CAAC,SAASF;oBAC1CG,cAAcd;oBACd,IAAIT,eAAeE,aAAaF;oBAChCI,QAAQ1B,KAAK;oBACb,MAAMW;oBACN,IAAImC;oBACJ,IAAI;wBACF,uFAAuF;wBACvF,uFAAuF;wBACvF,IAAI,CAACV,gBAAgB;4BACnB,MAAM,EAAEnC,OAAO8C,UAAU,EAAE,GAAG,MAAMzE,UAAUK;4BAC9C,MAAMoE,WAAW/C,KAAK;wBACxB;oBACF,EAAE,OAAOH,KAAK;wBACZiD,aAAajD;oBACf;oBACA,IAAI;wBACF,MAAME;oBACR,EAAE,OAAOF,KAAK;wBACZ,IAAIiD,YAAY,MAAM,IAAIE,eAAe;4BAACF;4BAAYjD;yBAAI,EAAE;wBAC5D,MAAMA;oBACR;oBACA,IAAIiD,YAAY,MAAMA;oBACtB,IAAIV,gBAAgB,MAAMA;gBAC5B,CAAA;gBACAG,aAAapB,IAAI,CAACe,iBAAiBC;YACrC;YACA,MAAMO,eAAe,IAAML;YAC3B3C,kBAAkB2C;YAClBG,QAAQS,IAAI,CAAC,UAAUP;YACvBF,QAAQS,IAAI,CAAC,WAAWP;YACxB,IAAIjD,cAAc4C,SAAS5C;iBACtB,KAAIb,eAAAA,KAAK+D,MAAM,cAAX/D,mCAAAA,aAAasE,OAAO,EAAEb;iBAC1B;oBACHzD;iBAAAA,gBAAAA,KAAK+D,MAAM,cAAX/D,oCAAAA,cAAauE,gBAAgB,CAAC,SAAST,cAAc;oBAAEO,MAAM;gBAAK;gBAClE,IAAI;oBACFpE,QAAQ;wBAAEkC,MAAM;wBAAWzB;wBAASa;oBAAO;oBAC3C,IAAI,CAACO,YAAaL,CAAAA,gBAAgBI,MAAM,GAAG,KAAKF,gBAAgB,CAAA,GAAI;wBAClE1B,QAAQ;4BAAEkC,MAAM;4BAAcT,QAAQC;4BAAeS,OAAOR;4BAAcJ,UAAUC;wBAAgB;oBACtG;gBACF,EAAE,OAAOR,KAAK;oBACZwC,SAASxC,eAAeuD,QAAQvD,MAAM,IAAIuD,MAAMC,OAAOxD;gBACzD;YACF;QACF;IACF,EAAE,OAAOA,KAAK;QACZ,IAAIC,mBAAmB,MAAMD;QAC7B,IAAI;YACF,MAAME;QACR,EAAE,OAAOuD,YAAY;YACnB,MAAM,IAAIN,eAAe;gBAACnD;gBAAKyD;aAAW,EAAE;QAC9C;QACA,MAAMzD;IACR,SAAU;QACRH,kBAAkB6D;IACpB;AACF"}
|
|
1
|
+
{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/watch.ts"],"sourcesContent":["import { watch as fsWatch } from 'node:fs';\nimport type { ResolvedConfig } from './config/index.ts';\nimport { SenseError } from './errors.ts';\nimport { guardedTick } from './lib/guarded-tick.ts';\nimport { shouldReconcileWatchNotification } from './lib/watch-notification.ts';\nimport { docCount, openStore } from './store/index.ts';\nimport { startWatchClaim, WATCH_HEARTBEAT_INTERVAL_MS } from './watch-claim.ts';\n\n// Watch is a cache pre-warmer, not a correctness mechanism: open() always reconciles anyway, so any fs event just triggers a debounced full reconcile.\nconst DEBOUNCE_MS = 200;\n\nexport type WatchEvent = { type: 'started'; rootDir: string; dbPath: string } | { type: 'reconciled'; parsed: number; total: number; warnings: string[] } | { type: 'reconcile-error'; message: string };\n\nexport interface WatchOptions {\n force?: boolean;\n onEvent?: (event: WatchEvent) => void;\n // Aborting runs the same shutdown path as SIGINT/SIGTERM.\n signal?: AbortSignal;\n debounceMs?: number;\n heartbeatIntervalMs?: number;\n}\n\n// Runs in the foreground until SIGINT/SIGTERM/signal abort. DuckDB and Turso hand the cache file to\n// one process at a time, so every reconcile cycle opens and closes instead of holding an idle lock.\nexport async function runWatch(cfg: ResolvedConfig, opts: WatchOptions = {}): Promise<void> {\n const onEvent = opts.onEvent ?? (() => {});\n const debounceMs = opts.debounceMs ?? DEBOUNCE_MS;\n const reconcileIntervalMs = opts.heartbeatIntervalMs ?? WATCH_HEARTBEAT_INTERVAL_MS;\n if (!Number.isFinite(reconcileIntervalMs) || reconcileIntervalMs <= 0) throw new SenseError('CONFIG_INVALID', 'watch heartbeatIntervalMs must be a finite positive number');\n // Reconciliation may be parked for a caller, but live ownership must renew before fixed expiry.\n const claimIntervalMs = Math.min(reconcileIntervalMs, WATCH_HEARTBEAT_INTERVAL_MS);\n const rootDir = cfg.rootDir ?? cfg.baseDir;\n const configDir = cfg.configDir ?? rootDir;\n let claimFailure: Error | null = null;\n let requestShutdown: ((reason: Error) => void) | undefined;\n const claim = await startWatchClaim(configDir, opts.force ?? false, claimIntervalMs, (err) => {\n claimFailure ??= err;\n requestShutdown?.(err);\n });\n let claimCloseStarted = false;\n const closeClaim = (): Promise<void> => {\n claimCloseStarted = true;\n return claim.close();\n };\n try {\n const { store: initialStore, dbPath, warnings: initialWarnings, parsed: initialParsed } = await openStore(cfg);\n let initialTotal = 0;\n try {\n // total is read before close: an event fires only once nothing is held open again.\n initialTotal = initialWarnings.length > 0 || initialParsed > 0 ? await docCount(initialStore) : 0;\n } finally {\n await initialStore.close();\n }\n if (claimFailure) throw claimFailure;\n\n let stopping = false;\n // At most one open+reconcile+close cycle runs at a time; a trigger that arrives mid-cycle is\n // coalesced into a single rerun instead of opening a second, overlapping connection.\n let running: Promise<void> | null = null;\n let queuedAlways: boolean | null = null;\n\n const tick = async (alwaysEmit: boolean): Promise<void> => {\n try {\n const { store, parsed, warnings } = await openStore(cfg);\n try {\n if (alwaysEmit || parsed > 0 || warnings.length > 0) {\n onEvent({ type: 'reconciled', parsed, total: await docCount(store), warnings });\n }\n } finally {\n await store.close();\n }\n } catch (err) {\n onEvent({ type: 'reconcile-error', message: (err as Error).message });\n }\n };\n\n const startCycle = (alwaysEmit: boolean) => {\n running = tick(alwaysEmit).then(() => {\n running = null;\n if (queuedAlways !== null && !stopping) {\n const again = queuedAlways;\n queuedAlways = null;\n startCycle(again);\n }\n });\n };\n\n const requestCycle = (alwaysEmit: boolean) => {\n if (stopping) return;\n if (running) {\n queuedAlways = queuedAlways || alwaysEmit;\n return;\n }\n startCycle(alwaysEmit);\n };\n\n let debounceTimer: NodeJS.Timeout | null = null;\n const scheduleReconcile = () => {\n if (debounceTimer) clearTimeout(debounceTimer);\n debounceTimer = setTimeout(() => {\n debounceTimer = null;\n requestCycle(true);\n }, debounceMs);\n };\n\n // Ignore our own state files, or cache and claim writes would retrigger reconciliation forever.\n // An unresolvable filename reconciles because parsing nothing costs less than missing a real edit.\n const watcher = fsWatch(rootDir, { recursive: true }, (_event, filename) => {\n if (!shouldReconcileWatchNotification(filename)) return;\n scheduleReconcile();\n });\n const reconcileTimer = setInterval(\n guardedTick(\n async () => requestCycle(false),\n () => stopping\n ),\n reconcileIntervalMs\n );\n\n return await new Promise<void>((resolveShutdown, rejectShutdown) => {\n let shutdownReason: Error | null = null;\n const shutdown = (reason?: Error) => {\n shutdownReason ??= reason ?? null;\n if (stopping) return;\n stopping = true;\n const shutdownTask = (async () => {\n process.off('SIGINT', stopNormally);\n process.off('SIGTERM', stopNormally);\n opts.signal?.removeEventListener('abort', stopNormally);\n clearInterval(reconcileTimer);\n if (debounceTimer) clearTimeout(debounceTimer);\n watcher.close();\n await running;\n let finalError: unknown;\n try {\n // A normal stop performs one final freshness pass. Ownership loss drains existing work\n // but starts nothing new because the replacement watcher now owns that responsibility.\n if (!shutdownReason) {\n const { store: finalStore } = await openStore(cfg);\n await finalStore.close();\n }\n } catch (err) {\n finalError = err;\n }\n try {\n await closeClaim();\n } catch (err) {\n if (finalError) throw new AggregateError([finalError, err], 'watch shutdown and claim cleanup both failed');\n throw err;\n }\n if (finalError) throw finalError;\n if (shutdownReason) throw shutdownReason;\n })();\n shutdownTask.then(resolveShutdown, rejectShutdown);\n };\n const stopNormally = () => shutdown();\n requestShutdown = shutdown;\n process.once('SIGINT', stopNormally);\n process.once('SIGTERM', stopNormally);\n if (claimFailure) shutdown(claimFailure);\n else if (opts.signal?.aborted) shutdown();\n else {\n opts.signal?.addEventListener('abort', stopNormally, { once: true });\n try {\n onEvent({ type: 'started', rootDir, dbPath });\n if (!stopping && (initialWarnings.length > 0 || initialParsed > 0)) {\n onEvent({ type: 'reconciled', parsed: initialParsed, total: initialTotal, warnings: initialWarnings });\n }\n } catch (err) {\n shutdown(err instanceof Error ? err : new Error(String(err)));\n }\n }\n });\n } catch (err) {\n if (claimCloseStarted) throw err;\n try {\n await closeClaim();\n } catch (closeError) {\n throw new AggregateError([err, closeError], 'watch setup and claim cleanup both failed');\n }\n throw err;\n } finally {\n requestShutdown = undefined;\n }\n}\n"],"names":["watch","fsWatch","SenseError","guardedTick","shouldReconcileWatchNotification","docCount","openStore","startWatchClaim","WATCH_HEARTBEAT_INTERVAL_MS","DEBOUNCE_MS","runWatch","cfg","opts","onEvent","debounceMs","reconcileIntervalMs","heartbeatIntervalMs","Number","isFinite","claimIntervalMs","Math","min","rootDir","baseDir","configDir","claimFailure","requestShutdown","claim","force","err","claimCloseStarted","closeClaim","close","store","initialStore","dbPath","warnings","initialWarnings","parsed","initialParsed","initialTotal","length","stopping","running","queuedAlways","tick","alwaysEmit","type","total","message","startCycle","then","again","requestCycle","debounceTimer","scheduleReconcile","clearTimeout","setTimeout","watcher","recursive","_event","filename","reconcileTimer","setInterval","Promise","resolveShutdown","rejectShutdown","shutdownReason","shutdown","reason","shutdownTask","process","off","stopNormally","signal","removeEventListener","clearInterval","finalError","finalStore","AggregateError","once","aborted","addEventListener","Error","String","closeError","undefined"],"mappings":"AAAA,SAASA,SAASC,OAAO,QAAQ,UAAU;AAE3C,SAASC,UAAU,QAAQ,cAAc;AACzC,SAASC,WAAW,QAAQ,wBAAwB;AACpD,SAASC,gCAAgC,QAAQ,8BAA8B;AAC/E,SAASC,QAAQ,EAAEC,SAAS,QAAQ,mBAAmB;AACvD,SAASC,eAAe,EAAEC,2BAA2B,QAAQ,mBAAmB;AAEhF,uJAAuJ;AACvJ,MAAMC,cAAc;AAapB,oGAAoG;AACpG,oGAAoG;AACpG,OAAO,eAAeC,SAASC,GAAmB,EAAEC,OAAqB,CAAC,CAAC;QACzDA,eACGA,kBACSA,2BAIZD,cACEA,gBAG6BC;IAV/C,MAAMC,WAAUD,gBAAAA,KAAKC,OAAO,cAAZD,2BAAAA,gBAAiB,KAAO;IACxC,MAAME,cAAaF,mBAAAA,KAAKE,UAAU,cAAfF,8BAAAA,mBAAmBH;IACtC,MAAMM,uBAAsBH,4BAAAA,KAAKI,mBAAmB,cAAxBJ,uCAAAA,4BAA4BJ;IACxD,IAAI,CAACS,OAAOC,QAAQ,CAACH,wBAAwBA,uBAAuB,GAAG,MAAM,IAAIb,WAAW,kBAAkB;IAC9G,gGAAgG;IAChG,MAAMiB,kBAAkBC,KAAKC,GAAG,CAACN,qBAAqBP;IACtD,MAAMc,WAAUX,eAAAA,IAAIW,OAAO,cAAXX,0BAAAA,eAAeA,IAAIY,OAAO;IAC1C,MAAMC,aAAYb,iBAAAA,IAAIa,SAAS,cAAbb,4BAAAA,iBAAiBW;IACnC,IAAIG,eAA6B;IACjC,IAAIC;IACJ,MAAMC,QAAQ,MAAMpB,gBAAgBiB,YAAWZ,cAAAA,KAAKgB,KAAK,cAAVhB,yBAAAA,cAAc,OAAOO,iBAAiB,CAACU;QACpFJ,yBAAAA,0BAAAA,eAAAA,eAAiBI;QACjBH,4BAAAA,sCAAAA,gBAAkBG;IACpB;IACA,IAAIC,oBAAoB;IACxB,MAAMC,aAAa;QACjBD,oBAAoB;QACpB,OAAOH,MAAMK,KAAK;IACpB;IACA,IAAI;QACF,MAAM,EAAEC,OAAOC,YAAY,EAAEC,MAAM,EAAEC,UAAUC,eAAe,EAAEC,QAAQC,aAAa,EAAE,GAAG,MAAMjC,UAAUK;QAC1G,IAAI6B,eAAe;QACnB,IAAI;YACF,mFAAmF;YACnFA,eAAeH,gBAAgBI,MAAM,GAAG,KAAKF,gBAAgB,IAAI,MAAMlC,SAAS6B,gBAAgB;QAClG,SAAU;YACR,MAAMA,aAAaF,KAAK;QAC1B;QACA,IAAIP,cAAc,MAAMA;QAExB,IAAIiB,WAAW;QACf,6FAA6F;QAC7F,qFAAqF;QACrF,IAAIC,UAAgC;QACpC,IAAIC,eAA+B;QAEnC,MAAMC,OAAO,OAAOC;YAClB,IAAI;gBACF,MAAM,EAAEb,KAAK,EAAEK,MAAM,EAAEF,QAAQ,EAAE,GAAG,MAAM9B,UAAUK;gBACpD,IAAI;oBACF,IAAImC,cAAcR,SAAS,KAAKF,SAASK,MAAM,GAAG,GAAG;wBACnD5B,QAAQ;4BAAEkC,MAAM;4BAAcT;4BAAQU,OAAO,MAAM3C,SAAS4B;4BAAQG;wBAAS;oBAC/E;gBACF,SAAU;oBACR,MAAMH,MAAMD,KAAK;gBACnB;YACF,EAAE,OAAOH,KAAK;gBACZhB,QAAQ;oBAAEkC,MAAM;oBAAmBE,SAAS,AAACpB,IAAcoB,OAAO;gBAAC;YACrE;QACF;QAEA,MAAMC,aAAa,CAACJ;YAClBH,UAAUE,KAAKC,YAAYK,IAAI,CAAC;gBAC9BR,UAAU;gBACV,IAAIC,iBAAiB,QAAQ,CAACF,UAAU;oBACtC,MAAMU,QAAQR;oBACdA,eAAe;oBACfM,WAAWE;gBACb;YACF;QACF;QAEA,MAAMC,eAAe,CAACP;YACpB,IAAIJ,UAAU;YACd,IAAIC,SAAS;gBACXC,eAAeA,gBAAgBE;gBAC/B;YACF;YACAI,WAAWJ;QACb;QAEA,IAAIQ,gBAAuC;QAC3C,MAAMC,oBAAoB;YACxB,IAAID,eAAeE,aAAaF;YAChCA,gBAAgBG,WAAW;gBACzBH,gBAAgB;gBAChBD,aAAa;YACf,GAAGvC;QACL;QAEA,gGAAgG;QAChG,mGAAmG;QACnG,MAAM4C,UAAUzD,QAAQqB,SAAS;YAAEqC,WAAW;QAAK,GAAG,CAACC,QAAQC;YAC7D,IAAI,CAACzD,iCAAiCyD,WAAW;YACjDN;QACF;QACA,MAAMO,iBAAiBC,YACrB5D,YACE,UAAYkD,aAAa,QACzB,IAAMX,WAER3B;QAGF,OAAO,MAAM,IAAIiD,QAAc,CAACC,iBAAiBC;gBAyCtCtD;YAxCT,IAAIuD,iBAA+B;YACnC,MAAMC,WAAW,CAACC;gBAChBF,2BAAAA,4BAAAA,iBAAAA,iBAAmBE,mBAAAA,oBAAAA,SAAU;gBAC7B,IAAI3B,UAAU;gBACdA,WAAW;gBACX,MAAM4B,eAAe,AAAC,CAAA;wBAGpB1D;oBAFA2D,QAAQC,GAAG,CAAC,UAAUC;oBACtBF,QAAQC,GAAG,CAAC,WAAWC;qBACvB7D,eAAAA,KAAK8D,MAAM,cAAX9D,mCAAAA,aAAa+D,mBAAmB,CAAC,SAASF;oBAC1CG,cAAcd;oBACd,IAAIR,eAAeE,aAAaF;oBAChCI,QAAQ1B,KAAK;oBACb,MAAMW;oBACN,IAAIkC;oBACJ,IAAI;wBACF,uFAAuF;wBACvF,uFAAuF;wBACvF,IAAI,CAACV,gBAAgB;4BACnB,MAAM,EAAElC,OAAO6C,UAAU,EAAE,GAAG,MAAMxE,UAAUK;4BAC9C,MAAMmE,WAAW9C,KAAK;wBACxB;oBACF,EAAE,OAAOH,KAAK;wBACZgD,aAAahD;oBACf;oBACA,IAAI;wBACF,MAAME;oBACR,EAAE,OAAOF,KAAK;wBACZ,IAAIgD,YAAY,MAAM,IAAIE,eAAe;4BAACF;4BAAYhD;yBAAI,EAAE;wBAC5D,MAAMA;oBACR;oBACA,IAAIgD,YAAY,MAAMA;oBACtB,IAAIV,gBAAgB,MAAMA;gBAC5B,CAAA;gBACAG,aAAanB,IAAI,CAACc,iBAAiBC;YACrC;YACA,MAAMO,eAAe,IAAML;YAC3B1C,kBAAkB0C;YAClBG,QAAQS,IAAI,CAAC,UAAUP;YACvBF,QAAQS,IAAI,CAAC,WAAWP;YACxB,IAAIhD,cAAc2C,SAAS3C;iBACtB,KAAIb,eAAAA,KAAK8D,MAAM,cAAX9D,mCAAAA,aAAaqE,OAAO,EAAEb;iBAC1B;oBACHxD;iBAAAA,gBAAAA,KAAK8D,MAAM,cAAX9D,oCAAAA,cAAasE,gBAAgB,CAAC,SAAST,cAAc;oBAAEO,MAAM;gBAAK;gBAClE,IAAI;oBACFnE,QAAQ;wBAAEkC,MAAM;wBAAWzB;wBAASa;oBAAO;oBAC3C,IAAI,CAACO,YAAaL,CAAAA,gBAAgBI,MAAM,GAAG,KAAKF,gBAAgB,CAAA,GAAI;wBAClE1B,QAAQ;4BAAEkC,MAAM;4BAAcT,QAAQC;4BAAeS,OAAOR;4BAAcJ,UAAUC;wBAAgB;oBACtG;gBACF,EAAE,OAAOR,KAAK;oBACZuC,SAASvC,eAAesD,QAAQtD,MAAM,IAAIsD,MAAMC,OAAOvD;gBACzD;YACF;QACF;IACF,EAAE,OAAOA,KAAK;QACZ,IAAIC,mBAAmB,MAAMD;QAC7B,IAAI;YACF,MAAME;QACR,EAAE,OAAOsD,YAAY;YACnB,MAAM,IAAIN,eAAe;gBAAClD;gBAAKwD;aAAW,EAAE;QAC9C;QACA,MAAMxD;IACR,SAAU;QACRH,kBAAkB4D;IACpB;AACF"}
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "sensemaking",
|
|
3
|
-
"version": "0.24.
|
|
3
|
+
"version": "0.24.7",
|
|
4
4
|
"description": "Query and search your markdown notes with context-aware progressive disclosure: SQL over frontmatter, links, and text, plus semantic search and link-graph ranking. No server, no build step.",
|
|
5
5
|
"keywords": [
|
|
6
6
|
"markdown",
|
|
@@ -93,7 +93,7 @@
|
|
|
93
93
|
"yaml": "^2.9.0"
|
|
94
94
|
},
|
|
95
95
|
"devDependencies": {
|
|
96
|
-
"@duckdb/node-api": "1.5.5-r.
|
|
96
|
+
"@duckdb/node-api": "1.5.5-r.5",
|
|
97
97
|
"@tursodatabase/database": "*",
|
|
98
98
|
"@types/markdown-it-footnote": "^3.0.4",
|
|
99
99
|
"@types/mocha": "*",
|
|
@@ -1,15 +1,15 @@
|
|
|
1
|
-
<!-- sense-store-benchmark release=0.24.
|
|
1
|
+
<!-- sense-store-benchmark release=0.24.7 -->
|
|
2
2
|
# Current store benchmark summary
|
|
3
3
|
|
|
4
|
-
The release assessment generated this file for store selection. Release `0.24.
|
|
4
|
+
The release assessment generated this file for store selection. Release `0.24.7` passed on 2026-09-17, measured on Apple M4 Pro with Node v26.7.0.
|
|
5
5
|
|
|
6
6
|
The timing rows ran on the same 6,566-note tree. They include CLI startup and each store's complete selected path. Ranked candidates and downstream work can differ by store, so these are current operating measurements rather than an isolated database-engine contest.
|
|
7
7
|
|
|
8
8
|
| Store | Cold index | Warm count | Lexical search | Semantic search | Portable semantic nDCG@10 |
|
|
9
9
|
|---|---|---|---|---|---|
|
|
10
|
-
| sqlite | 1,
|
|
11
|
-
| duckdb |
|
|
12
|
-
| turso |
|
|
10
|
+
| sqlite | 1,641 ms | 156 ms | 210 ms | 349 ms | 0.3306 |
|
|
11
|
+
| duckdb | 3,376 ms | 181 ms | 565 ms | 556 ms | 0.3284 |
|
|
12
|
+
| turso | 3,704 ms | 154 ms | 296 ms | 664 ms | 0.3307 |
|
|
13
13
|
|
|
14
14
|
Cold index is the first `status` that builds the cache. Warm count is a no-change `COUNT(*)` query. Lexical and semantic search are steady-state `sense search` commands. Lower timing is faster. Higher nDCG@10 is better; that quality column uses the same NFCorpus queries, judgments, result count, and model on every store.
|
|
15
15
|
|