@cerefox/memory 1.1.0-beta.2 → 1.1.0-beta.4

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.
@@ -7438,9 +7438,143 @@ var exports_meta = {};
7438
7438
  __export(exports_meta, {
7439
7439
  PKG_VERSION: () => PKG_VERSION
7440
7440
  });
7441
- var PKG_VERSION = "1.1.0-beta.2";
7441
+ var PKG_VERSION = "1.1.0-beta.4";
7442
7442
  var init_meta = () => {};
7443
7443
 
7444
+ // ../../_shared/config/paths.ts
7445
+ import { existsSync, readFileSync } from "node:fs";
7446
+ import { homedir } from "node:os";
7447
+ import { join, resolve as resolvePath } from "node:path";
7448
+ import { cwd as processCwd, env } from "node:process";
7449
+ function expandTilde(p) {
7450
+ if (p === "~" || p.startsWith("~/")) {
7451
+ return join(homedir(), p.slice(2));
7452
+ }
7453
+ return p;
7454
+ }
7455
+ function userStateDirAbs(opts) {
7456
+ return resolvePath(join(opts.home ?? homedir(), USER_STATE_DIR_NAME));
7457
+ }
7458
+ function resolveConfigDir(opts = {}) {
7459
+ const override = (env.CEREFOX_CONFIG_DIR ?? "").trim();
7460
+ if (override) {
7461
+ return resolvePath(expandTilde(override));
7462
+ }
7463
+ const userState = userStateDirAbs(opts);
7464
+ if (existsSync(join(userState, ".env"))) {
7465
+ return userState;
7466
+ }
7467
+ const here = opts.cwd ?? processCwd();
7468
+ const cwdEnv = join(here, ".env");
7469
+ if (existsSync(cwdEnv) && cwdEnvHasCerefoxKey(cwdEnv)) {
7470
+ return resolvePath(here);
7471
+ }
7472
+ return userState;
7473
+ }
7474
+ function cwdEnvHasCerefoxKey(envPath) {
7475
+ try {
7476
+ const contents = readFileSync(envPath, "utf8");
7477
+ return /^\s*CEREFOX_[A-Z0-9_]+\s*=/m.test(contents);
7478
+ } catch {
7479
+ return false;
7480
+ }
7481
+ }
7482
+ function resolveEnvFile(opts = {}) {
7483
+ return join(resolveConfigDir(opts), ".env");
7484
+ }
7485
+ function userStateDir(opts = {}) {
7486
+ return userStateDirAbs(opts);
7487
+ }
7488
+ function isDevMode(opts = {}) {
7489
+ if ((env.CEREFOX_CONFIG_DIR ?? "").trim())
7490
+ return false;
7491
+ if (existsSync(join(userStateDirAbs(opts), ".env")))
7492
+ return false;
7493
+ const here = opts.cwd ?? processCwd();
7494
+ const cwdEnv = join(here, ".env");
7495
+ return existsSync(cwdEnv) && cwdEnvHasCerefoxKey(cwdEnv);
7496
+ }
7497
+ var USER_STATE_DIR_NAME = ".cerefox";
7498
+ var init_paths = () => {};
7499
+
7500
+ // ../../_shared/config/env.ts
7501
+ import { readFileSync as readFileSync2 } from "node:fs";
7502
+ import { env as env2 } from "node:process";
7503
+ function parseDotenv(content) {
7504
+ const result = {};
7505
+ for (const rawLine of content.split(/\r?\n/)) {
7506
+ const line = rawLine.trim();
7507
+ if (!line || line.startsWith("#"))
7508
+ continue;
7509
+ const match = line.match(KV_LINE);
7510
+ if (!match)
7511
+ continue;
7512
+ let value = match[2];
7513
+ if (value.length >= 2 && (value.startsWith('"') && value.endsWith('"') || value.startsWith("'") && value.endsWith("'"))) {
7514
+ value = value.slice(1, -1);
7515
+ }
7516
+ result[match[1]] = value;
7517
+ }
7518
+ return result;
7519
+ }
7520
+ function loadEnv(opts = {}) {
7521
+ if (_loaded) {
7522
+ return { path: "(already loaded)", vars: 0 };
7523
+ }
7524
+ _loaded = true;
7525
+ const envPath = resolveEnvFile(opts);
7526
+ let content;
7527
+ try {
7528
+ content = readFileSync2(envPath, "utf8");
7529
+ } catch {
7530
+ return { path: envPath, vars: 0 };
7531
+ }
7532
+ const configDirNamed = (env2.CEREFOX_CONFIG_DIR ?? "").trim() !== "";
7533
+ let count = 0;
7534
+ for (const [k, v] of Object.entries(parseDotenv(content))) {
7535
+ if (k === "CEREFOX_CONFIG_DIR")
7536
+ continue;
7537
+ if (env2[k] === undefined || configDirNamed) {
7538
+ env2[k] = v;
7539
+ count++;
7540
+ }
7541
+ }
7542
+ return { path: envPath, vars: count };
7543
+ }
7544
+ function loadSettings(opts = {}) {
7545
+ loadEnv(opts);
7546
+ return {
7547
+ supabaseUrl: env2.CEREFOX_SUPABASE_URL ?? "",
7548
+ supabaseKey: env2.CEREFOX_SUPABASE_KEY ?? "",
7549
+ supabaseAnonKey: env2.CEREFOX_SUPABASE_ANON_KEY ?? "",
7550
+ accessToken: env2.CEREFOX_ACCESS_TOKEN ?? "",
7551
+ databaseUrl: env2.CEREFOX_DATABASE_URL ?? "",
7552
+ openaiApiKey: env2.CEREFOX_OPENAI_API_KEY ?? env2.OPENAI_API_KEY ?? "",
7553
+ fireworksApiKey: env2.CEREFOX_FIREWORKS_API_KEY ?? ""
7554
+ };
7555
+ }
7556
+ var KV_LINE, _loaded = false;
7557
+ var init_env = __esm(() => {
7558
+ init_paths();
7559
+ KV_LINE = /^\s*([A-Za-z_][A-Za-z0-9_]*)\s*=\s*(.*?)\s*$/;
7560
+ });
7561
+
7562
+ // ../../_shared/config/index.ts
7563
+ var exports_config = {};
7564
+ __export(exports_config, {
7565
+ userStateDir: () => userStateDir,
7566
+ resolveEnvFile: () => resolveEnvFile,
7567
+ resolveConfigDir: () => resolveConfigDir,
7568
+ loadSettings: () => loadSettings,
7569
+ loadEnv: () => loadEnv,
7570
+ isDevMode: () => isDevMode,
7571
+ USER_STATE_DIR_NAME: () => USER_STATE_DIR_NAME
7572
+ });
7573
+ var init_config = __esm(() => {
7574
+ init_paths();
7575
+ init_env();
7576
+ });
7577
+
7444
7578
  // ../../node_modules/.bun/tslib@2.8.1/node_modules/tslib/tslib.js
7445
7579
  var require_tslib = __commonJS((exports, module) => {
7446
7580
  var __extends;
@@ -7935,7 +8069,7 @@ var require_tslib = __commonJS((exports, module) => {
7935
8069
  throw new TypeError("Cannot use 'in' operator on non-object");
7936
8070
  return typeof state === "function" ? receiver === state : state.has(receiver);
7937
8071
  };
7938
- __addDisposableResource = function(env, value, async) {
8072
+ __addDisposableResource = function(env3, value, async) {
7939
8073
  if (value !== null && value !== undefined) {
7940
8074
  if (typeof value !== "object" && typeof value !== "function")
7941
8075
  throw new TypeError("Object expected.");
@@ -7962,9 +8096,9 @@ var require_tslib = __commonJS((exports, module) => {
7962
8096
  return Promise.reject(e);
7963
8097
  }
7964
8098
  };
7965
- env.stack.push({ value, dispose, async });
8099
+ env3.stack.push({ value, dispose, async });
7966
8100
  } else if (async) {
7967
- env.stack.push({ async: true });
8101
+ env3.stack.push({ async: true });
7968
8102
  }
7969
8103
  return value;
7970
8104
  };
@@ -7972,17 +8106,17 @@ var require_tslib = __commonJS((exports, module) => {
7972
8106
  var e = new Error(message);
7973
8107
  return e.name = "SuppressedError", e.error = error, e.suppressed = suppressed, e;
7974
8108
  };
7975
- __disposeResources = function(env) {
8109
+ __disposeResources = function(env3) {
7976
8110
  function fail(e) {
7977
- env.error = env.hasError ? new _SuppressedError(e, env.error, "An error was suppressed during disposal.") : e;
7978
- env.hasError = true;
8111
+ env3.error = env3.hasError ? new _SuppressedError(e, env3.error, "An error was suppressed during disposal.") : e;
8112
+ env3.hasError = true;
7979
8113
  }
7980
8114
  var r, s = 0;
7981
8115
  function next() {
7982
- while (r = env.stack.pop()) {
8116
+ while (r = env3.stack.pop()) {
7983
8117
  try {
7984
8118
  if (!r.async && s === 1)
7985
- return s = 0, env.stack.push(r), Promise.resolve().then(next);
8119
+ return s = 0, env3.stack.push(r), Promise.resolve().then(next);
7986
8120
  if (r.dispose) {
7987
8121
  var result = r.dispose.call(r.value);
7988
8122
  if (r.async)
@@ -7997,9 +8131,9 @@ var require_tslib = __commonJS((exports, module) => {
7997
8131
  }
7998
8132
  }
7999
8133
  if (s === 1)
8000
- return env.hasError ? Promise.reject(env.error) : Promise.resolve();
8001
- if (env.hasError)
8002
- throw env.error;
8134
+ return env3.hasError ? Promise.reject(env3.error) : Promise.resolve();
8135
+ if (env3.hasError)
8136
+ throw env3.error;
8003
8137
  }
8004
8138
  return next();
8005
8139
  };
@@ -9128,22 +9262,22 @@ var require_websocket_factory = __commonJS((exports) => {
9128
9262
  };
9129
9263
  }
9130
9264
  static getWebSocketConstructor() {
9131
- const env = this.detectEnvironment();
9132
- if (env.wsConstructor) {
9133
- return env.wsConstructor;
9265
+ const env3 = this.detectEnvironment();
9266
+ if (env3.wsConstructor) {
9267
+ return env3.wsConstructor;
9134
9268
  }
9135
- let errorMessage = env.error || "WebSocket not supported in this environment.";
9136
- if (env.workaround) {
9269
+ let errorMessage = env3.error || "WebSocket not supported in this environment.";
9270
+ if (env3.workaround) {
9137
9271
  errorMessage += `
9138
9272
 
9139
- Suggested solution: ${env.workaround}`;
9273
+ Suggested solution: ${env3.workaround}`;
9140
9274
  }
9141
9275
  throw new Error(errorMessage);
9142
9276
  }
9143
9277
  static isWebSocketSupported() {
9144
9278
  try {
9145
- const env = this.detectEnvironment();
9146
- return env.type === "native";
9279
+ const env3 = this.detectEnvironment();
9280
+ return env3.type === "native";
9147
9281
  } catch (_a) {
9148
9282
  return false;
9149
9283
  }
@@ -11592,8 +11726,8 @@ var require_RealtimeChannel = __commonJS((exports) => {
11592
11726
  });
11593
11727
  }
11594
11728
  _notThisChannelEvent(event, ref) {
11595
- const { close, error, leave, join } = constants_1.CHANNEL_EVENTS;
11596
- const events = [close, error, leave, join];
11729
+ const { close, error, leave, join: join2 } = constants_1.CHANNEL_EVENTS;
11730
+ const events = [close, error, leave, join2];
11597
11731
  return ref && events.includes(event) && ref !== this.joinPush.ref;
11598
11732
  }
11599
11733
  _updateFilterTransform() {
@@ -23274,140 +23408,6 @@ var init_db_client = __esm(() => {
23274
23408
  });
23275
23409
  });
23276
23410
 
23277
- // ../../_shared/config/paths.ts
23278
- import { existsSync, readFileSync } from "node:fs";
23279
- import { homedir } from "node:os";
23280
- import { join, resolve as resolvePath } from "node:path";
23281
- import { cwd as processCwd, env } from "node:process";
23282
- function expandTilde(p) {
23283
- if (p === "~" || p.startsWith("~/")) {
23284
- return join(homedir(), p.slice(2));
23285
- }
23286
- return p;
23287
- }
23288
- function userStateDirAbs(opts) {
23289
- return resolvePath(join(opts.home ?? homedir(), USER_STATE_DIR_NAME));
23290
- }
23291
- function resolveConfigDir(opts = {}) {
23292
- const override = (env.CEREFOX_CONFIG_DIR ?? "").trim();
23293
- if (override) {
23294
- return resolvePath(expandTilde(override));
23295
- }
23296
- const userState = userStateDirAbs(opts);
23297
- if (existsSync(join(userState, ".env"))) {
23298
- return userState;
23299
- }
23300
- const here = opts.cwd ?? processCwd();
23301
- const cwdEnv = join(here, ".env");
23302
- if (existsSync(cwdEnv) && cwdEnvHasCerefoxKey(cwdEnv)) {
23303
- return resolvePath(here);
23304
- }
23305
- return userState;
23306
- }
23307
- function cwdEnvHasCerefoxKey(envPath) {
23308
- try {
23309
- const contents = readFileSync(envPath, "utf8");
23310
- return /^\s*CEREFOX_[A-Z0-9_]+\s*=/m.test(contents);
23311
- } catch {
23312
- return false;
23313
- }
23314
- }
23315
- function resolveEnvFile(opts = {}) {
23316
- return join(resolveConfigDir(opts), ".env");
23317
- }
23318
- function userStateDir(opts = {}) {
23319
- return userStateDirAbs(opts);
23320
- }
23321
- function isDevMode(opts = {}) {
23322
- if ((env.CEREFOX_CONFIG_DIR ?? "").trim())
23323
- return false;
23324
- if (existsSync(join(userStateDirAbs(opts), ".env")))
23325
- return false;
23326
- const here = opts.cwd ?? processCwd();
23327
- const cwdEnv = join(here, ".env");
23328
- return existsSync(cwdEnv) && cwdEnvHasCerefoxKey(cwdEnv);
23329
- }
23330
- var USER_STATE_DIR_NAME = ".cerefox";
23331
- var init_paths = () => {};
23332
-
23333
- // ../../_shared/config/env.ts
23334
- import { readFileSync as readFileSync2 } from "node:fs";
23335
- import { env as env2 } from "node:process";
23336
- function parseDotenv(content) {
23337
- const result = {};
23338
- for (const rawLine of content.split(/\r?\n/)) {
23339
- const line = rawLine.trim();
23340
- if (!line || line.startsWith("#"))
23341
- continue;
23342
- const match = line.match(KV_LINE);
23343
- if (!match)
23344
- continue;
23345
- let value = match[2];
23346
- if (value.length >= 2 && (value.startsWith('"') && value.endsWith('"') || value.startsWith("'") && value.endsWith("'"))) {
23347
- value = value.slice(1, -1);
23348
- }
23349
- result[match[1]] = value;
23350
- }
23351
- return result;
23352
- }
23353
- function loadEnv(opts = {}) {
23354
- if (_loaded) {
23355
- return { path: "(already loaded)", vars: 0 };
23356
- }
23357
- _loaded = true;
23358
- const envPath = resolveEnvFile(opts);
23359
- let content;
23360
- try {
23361
- content = readFileSync2(envPath, "utf8");
23362
- } catch {
23363
- return { path: envPath, vars: 0 };
23364
- }
23365
- const configDirNamed = (env2.CEREFOX_CONFIG_DIR ?? "").trim() !== "";
23366
- let count = 0;
23367
- for (const [k, v] of Object.entries(parseDotenv(content))) {
23368
- if (k === "CEREFOX_CONFIG_DIR")
23369
- continue;
23370
- if (env2[k] === undefined || configDirNamed) {
23371
- env2[k] = v;
23372
- count++;
23373
- }
23374
- }
23375
- return { path: envPath, vars: count };
23376
- }
23377
- function loadSettings(opts = {}) {
23378
- loadEnv(opts);
23379
- return {
23380
- supabaseUrl: env2.CEREFOX_SUPABASE_URL ?? "",
23381
- supabaseKey: env2.CEREFOX_SUPABASE_KEY ?? "",
23382
- supabaseAnonKey: env2.CEREFOX_SUPABASE_ANON_KEY ?? "",
23383
- accessToken: env2.CEREFOX_ACCESS_TOKEN ?? "",
23384
- databaseUrl: env2.CEREFOX_DATABASE_URL ?? "",
23385
- openaiApiKey: env2.CEREFOX_OPENAI_API_KEY ?? env2.OPENAI_API_KEY ?? "",
23386
- fireworksApiKey: env2.CEREFOX_FIREWORKS_API_KEY ?? ""
23387
- };
23388
- }
23389
- var KV_LINE, _loaded = false;
23390
- var init_env = __esm(() => {
23391
- init_paths();
23392
- KV_LINE = /^\s*([A-Za-z_][A-Za-z0-9_]*)\s*=\s*(.*?)\s*$/;
23393
- });
23394
-
23395
- // ../../_shared/config/index.ts
23396
- var exports_config = {};
23397
- __export(exports_config, {
23398
- userStateDir: () => userStateDir,
23399
- resolveEnvFile: () => resolveEnvFile,
23400
- resolveConfigDir: () => resolveConfigDir,
23401
- loadSettings: () => loadSettings,
23402
- loadEnv: () => loadEnv,
23403
- isDevMode: () => isDevMode,
23404
- USER_STATE_DIR_NAME: () => USER_STATE_DIR_NAME
23405
- });
23406
- var init_config = __esm(() => {
23407
- init_paths();
23408
- init_env();
23409
- });
23410
-
23411
23411
  // src/cli/util/client.ts
23412
23412
  function getClient() {
23413
23413
  const settings = loadSettings();
@@ -69592,6 +69592,7 @@ init_cli_core();
69592
69592
 
69593
69593
  // src/cli/commands/backup.ts
69594
69594
  init_cli_core();
69595
+ init_config();
69595
69596
  import { existsSync as existsSync2, mkdirSync, writeFileSync } from "node:fs";
69596
69597
  import { homedir as homedir2 } from "node:os";
69597
69598
  import { join as join2, resolve } from "node:path";
@@ -69635,6 +69636,7 @@ function utcStamp() {
69635
69636
  return d.getUTCFullYear().toString() + pad(d.getUTCMonth() + 1) + pad(d.getUTCDate()) + "T" + pad(d.getUTCHours()) + pad(d.getUTCMinutes()) + pad(d.getUTCSeconds()) + "Z";
69636
69637
  }
69637
69638
  async function action(options) {
69639
+ loadEnv();
69638
69640
  const configuredDir = options.outputDir ?? process.env.CEREFOX_BACKUP_DIR;
69639
69641
  const outDir = resolve(expandHome(configuredDir ?? "~/.cerefox/backups"));
69640
69642
  if (configuredDir !== undefined && !configuredDir.startsWith("/") && !configuredDir.startsWith("~")) {
@@ -69644,8 +69646,9 @@ async function action(options) {
69644
69646
  }
69645
69647
  if (!existsSync2(outDir))
69646
69648
  mkdirSync(outDir, { recursive: true });
69649
+ const envLabel = (process.env.CEREFOX_ENV_LABEL ?? "").trim().replace(/[^A-Za-z0-9._-]+/g, "-").replace(/^-+|-+$/g, "");
69647
69650
  const stamp = utcStamp();
69648
- const filename = `cerefox-${stamp}${options.label ? "-" + options.label : ""}.json`;
69651
+ const filename = `cerefox-${envLabel ? envLabel + "-" : ""}${stamp}` + `${options.label ? "-" + options.label : ""}.json`;
69649
69652
  const dest = join2(outDir, filename);
69650
69653
  const client = getClient();
69651
69654
  let schemaVersion = "unknown";
@@ -69725,6 +69728,7 @@ async function action(options) {
69725
69728
  document_count: docs.length,
69726
69729
  trashed_count: trashedCount,
69727
69730
  includes_trash: includeTrash,
69731
+ env_label: envLabel || null,
69728
69732
  includes_lifecycle_status: lifecycleCaptured,
69729
69733
  chunk_count: chunkTotal,
69730
69734
  project_count: projects.length,
@@ -76226,7 +76230,8 @@ async function action15(options) {
76226
76230
  if (options.json) {
76227
76231
  printJson(results);
76228
76232
  } else {
76229
- println("Cerefox doctor");
76233
+ const envLabel = (process.env.CEREFOX_ENV_LABEL ?? "").trim();
76234
+ println(envLabel ? `Cerefox doctor ${c.yellow(`[${envLabel.toUpperCase()}]`)}` : "Cerefox doctor");
76230
76235
  println("");
76231
76236
  const nameWidth = Math.max(...results.map((r) => r.name.length));
76232
76237
  for (const r of results) {
@@ -76788,7 +76793,8 @@ class IngestionPipeline {
76788
76793
  author = "unknown",
76789
76794
  authorType = "user",
76790
76795
  expectedContentHash,
76791
- lastWriteWins = false
76796
+ lastWriteWins = false,
76797
+ forceRechunk = false
76792
76798
  } = opts;
76793
76799
  const listFormProvided = projectIds !== undefined && projectIds !== null || projectNames !== undefined && projectNames !== null;
76794
76800
  const getOrCreate = (name) => this.db.getOrCreateProject(name);
@@ -76811,7 +76817,8 @@ class IngestionPipeline {
76811
76817
  author,
76812
76818
  authorType,
76813
76819
  expectedContentHash,
76814
- lastWriteWins
76820
+ lastWriteWins,
76821
+ forceRechunk
76815
76822
  });
76816
76823
  if (!listFormProvided && (projectId || projectName)) {
76817
76824
  const singular = await resolveProjectIds({ projectId, projectName }, getOrCreate);
@@ -76936,7 +76943,8 @@ class IngestionPipeline {
76936
76943
  author = "unknown",
76937
76944
  authorType = "user",
76938
76945
  expectedContentHash,
76939
- lastWriteWins = false
76946
+ lastWriteWins = false,
76947
+ forceRechunk = false
76940
76948
  } = opts;
76941
76949
  const existing = await this.db.getDocumentById(documentId);
76942
76950
  if (!existing) {
@@ -76961,7 +76969,7 @@ class IngestionPipeline {
76961
76969
  }
76962
76970
  const actualChunks = await this.db.listChunksForDocument(documentId);
76963
76971
  const hasChunks = actualChunks.length > 0;
76964
- if (contentUnchanged && hasChunks) {
76972
+ if (contentUnchanged && hasChunks && !forceRechunk) {
76965
76973
  const oldTitle = existing.title ?? "";
76966
76974
  const titleChanged = oldTitle !== title;
76967
76975
  const updates = { title };
@@ -77048,7 +77056,7 @@ class IngestionPipeline {
77048
77056
  sourceLabel: source,
77049
77057
  retentionHours: this.settings.versionRetentionHours,
77050
77058
  cleanupEnabled: this.settings.versionCleanupEnabled,
77051
- expectedContentHash: expectedContentHash ?? null,
77059
+ expectedContentHash: expectedContentHash ?? (forceRechunk && contentUnchanged ? existing.content_hash : null),
77052
77060
  lastWriteWins
77053
77061
  });
77054
77062
  let finalProjectIds;
@@ -78027,6 +78035,25 @@ function registerMetadataSearch(program2) {
78027
78035
  init_dist4();
78028
78036
  init_cli_core();
78029
78037
  init_config();
78038
+
78039
+ // src/cli/util/bulk-write-warning.ts
78040
+ init_cli_core();
78041
+ function warnLargeBulkWrite(opts) {
78042
+ if (opts.count < opts.threshold)
78043
+ return false;
78044
+ println("");
78045
+ println(c.yellow("⚠ ") + `${opts.count.toLocaleString()} ${opts.unit}(s) is a large bulk rewrite.`);
78046
+ println(c.dim(` On Supabase (especially the free tier and small compute add-ons) this can
78047
+ ` + ` deplete the project's Disk IO Budget. Symptoms: slower responses, CPU
78048
+ ` + ` climbing on IO wait, and in the worst case a briefly unresponsive
78049
+ ` + " instance. It recovers on its own once the budget refills."));
78050
+ println(c.dim(` Gentler: ${opts.batchHint}, leaving time between runs and ideally
78051
+ ` + " picking a quiet period. The work is resumable — re-run to continue."));
78052
+ println("");
78053
+ return true;
78054
+ }
78055
+
78056
+ // src/cli/commands/reindex.ts
78030
78057
  async function action27(options) {
78031
78058
  const settings = loadSettings();
78032
78059
  if (!settings.supabaseUrl || !settings.supabaseKey) {
@@ -78065,6 +78092,12 @@ async function action27(options) {
78065
78092
  return;
78066
78093
  }
78067
78094
  println(c.bold(`Reindexing ${chunks.length} chunk(s) ${reindexAll ? "(--all)" : "(stale only)"}${dryRun ? " — DRY RUN" : ""}`));
78095
+ warnLargeBulkWrite({
78096
+ count: chunks.length,
78097
+ threshold: 1000,
78098
+ unit: "chunk",
78099
+ batchHint: "reindex in stages with --document-id"
78100
+ });
78068
78101
  if (dryRun) {
78069
78102
  const byDoc = new Map;
78070
78103
  for (const ch of chunks)
@@ -78134,7 +78167,16 @@ async function action28(options) {
78134
78167
  } catch (err) {
78135
78168
  throw systemError(`Could not list legacy chunks: ${err instanceof Error ? err.message : String(err)}`);
78136
78169
  }
78137
- const docIds = [...new Set(legacyChunkRows.map((r) => r.document_id))];
78170
+ let liveIds = new Set;
78171
+ try {
78172
+ const liveRows = await fetchAllPages((from, to) => supabase.from("cerefox_documents").select("id").is("deleted_at", null).order("id", { ascending: true }).range(from, to));
78173
+ liveIds = new Set(liveRows.map((r) => r.id));
78174
+ } catch (err) {
78175
+ throw systemError(`Could not list documents: ${err instanceof Error ? err.message : String(err)}`);
78176
+ }
78177
+ const allLegacy = [...new Set(legacyChunkRows.map((r) => r.document_id))];
78178
+ const docIds = allLegacy.filter((id) => liveIds.has(id));
78179
+ const trashedSkipped = allLegacy.length - docIds.length;
78138
78180
  const limit = options.limit ? parsePositiveInt(options.limit, "--limit", docIds.length) : docIds.length;
78139
78181
  const targets = docIds.slice(0, limit);
78140
78182
  if (targets.length === 0) {
@@ -78142,6 +78184,15 @@ async function action28(options) {
78142
78184
  return;
78143
78185
  }
78144
78186
  println(c.bold(`${docIds.length} document(s) on the legacy format` + (targets.length < docIds.length ? `; converting ${targets.length} (--limit)` : "")));
78187
+ if (trashedSkipped > 0) {
78188
+ println(c.dim(` (${trashedSkipped} trashed document(s) on the legacy format ignored)`));
78189
+ }
78190
+ warnLargeBulkWrite({
78191
+ count: targets.length,
78192
+ threshold: 500,
78193
+ unit: "document",
78194
+ batchHint: "run it in batches with --limit 200"
78195
+ });
78145
78196
  if (options.dryRun) {
78146
78197
  println(c.yellow("⚠ --dry-run: nothing was written."));
78147
78198
  println(c.dim(" Each document would be re-chunked and RE-EMBEDDED (embedding spend)."));
@@ -78177,16 +78228,24 @@ async function action28(options) {
78177
78228
  continue;
78178
78229
  }
78179
78230
  try {
78180
- await pipeline2.ingestText({
78231
+ const result = await pipeline2.ingestText({
78181
78232
  text: doc2.full_content,
78182
78233
  title: doc2.doc_title,
78183
78234
  documentId: id,
78184
78235
  source: "migrate-format",
78185
78236
  author,
78186
78237
  authorType,
78187
- expectedContentHash: doc2.content_hash
78238
+ expectedContentHash: doc2.content_hash,
78239
+ forceRechunk: true
78188
78240
  });
78189
- converted++;
78241
+ if (result.reindexed) {
78242
+ converted++;
78243
+ } else {
78244
+ failures.push({
78245
+ document: `${doc2.doc_title} (${id})`,
78246
+ reason: `pipeline reported no re-chunk (action=${result.action}); format not advanced`
78247
+ });
78248
+ }
78190
78249
  } catch (err) {
78191
78250
  const message = err instanceof Error ? err.message : String(err);
78192
78251
  if (/conflict/i.test(message)) {
@@ -78268,6 +78327,13 @@ async function action29(target, options) {
78268
78327
  } else {
78269
78328
  warn("This backup predates project-membership capture (format 1) — documents " + "will be restored WITHOUT their project assignments.");
78270
78329
  }
78330
+ {
78331
+ const fileEnv = (payload.env_label ?? "").trim();
78332
+ const targetEnv = (process.env.CEREFOX_ENV_LABEL ?? "").trim();
78333
+ if (fileEnv !== targetEnv) {
78334
+ warn(`This snapshot came from ${fileEnv ? `the "${fileEnv}" environment` : "an unlabelled (production) environment"}, ` + `but you are restoring into ${targetEnv ? `"${targetEnv}"` : "an unlabelled (production) environment"}.`);
78335
+ }
78336
+ }
78271
78337
  const trashedInFile = payload.documents.filter((d) => d.deleted_at != null).length;
78272
78338
  if (trashedInFile > 0) {
78273
78339
  println(c.dim(` trashed documents in file: ${trashedInFile} — restored as trash ` + "(still deleted; recover with `cerefox document restore`)"));
@@ -83600,7 +83666,10 @@ var VERSION_INFO = {
83600
83666
  };
83601
83667
  var SCHEMA_VERSION_RE3 = /^--\s*@version:\s*(\S+)/m;
83602
83668
  function registerMetaRoutes(app, ctx) {
83603
- app.get("/api/v1/version", (c2) => c2.json(VERSION_INFO));
83669
+ app.get("/api/v1/version", (c2) => {
83670
+ const label = (process.env.CEREFOX_ENV_LABEL ?? "").trim();
83671
+ return c2.json({ ...VERSION_INFO, env_label: label.length > 0 ? label : null });
83672
+ });
83604
83673
  app.get("/api/v1/docs", (c2) => c2.json(listBundledDocs2()));
83605
83674
  app.get("/api/v1/docs/:path{.+}", (c2) => {
83606
83675
  const docPath = c2.req.param("path");
@@ -84426,6 +84495,12 @@ async function bareEntryPoint() {
84426
84495
  }
84427
84496
  }
84428
84497
  async function main() {
84498
+ {
84499
+ const { loadEnv: loadEnv2 } = await Promise.resolve().then(() => (init_config(), exports_config));
84500
+ try {
84501
+ loadEnv2();
84502
+ } catch {}
84503
+ }
84429
84504
  if (process.argv.length === 2) {
84430
84505
  await bareEntryPoint();
84431
84506
  return;