bluera-knowledge 0.19.7 → 0.21.0

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.
@@ -229,6 +229,12 @@ declare class CodeGraphService {
229
229
  interface EmbeddingConfig {
230
230
  readonly model: string;
231
231
  readonly batchSize: number;
232
+ readonly dtype: 'fp32' | 'fp16' | 'q8' | 'q4';
233
+ readonly pooling: 'mean' | 'cls' | 'none';
234
+ readonly normalize: boolean;
235
+ readonly queryPrefix: string;
236
+ readonly docPrefix: string;
237
+ readonly maxInFlightBatches: number;
232
238
  }
233
239
  interface IndexingConfig {
234
240
  readonly concurrency: number;
@@ -343,9 +349,8 @@ declare class EmbeddingEngine {
343
349
  private initPromise;
344
350
  private _dimensions;
345
351
  private disposed;
346
- private readonly modelName;
347
- private readonly batchSize;
348
- constructor(modelName?: string, batchSize?: number);
352
+ private readonly config;
353
+ constructor(config?: EmbeddingConfig);
349
354
  /**
350
355
  * Guard against use-after-dispose
351
356
  */
@@ -355,13 +360,53 @@ declare class EmbeddingEngine {
355
360
  * Multiple concurrent calls will share the same initialization promise.
356
361
  */
357
362
  initialize(): Promise<void>;
358
- embed(text: string): Promise<number[]>;
359
- embedBatch(texts: string[]): Promise<number[][]>;
363
+ /**
364
+ * Embed a search query. Applies queryPrefix for asymmetric models.
365
+ */
366
+ embedQuery(text: string): Promise<Float32Array>;
367
+ /**
368
+ * Embed a document for indexing. Applies docPrefix for asymmetric models.
369
+ */
370
+ embedDocument(text: string): Promise<Float32Array>;
371
+ /**
372
+ * Internal: embed text without prefix.
373
+ */
374
+ private embedText;
375
+ /**
376
+ * Embed a batch of documents with optional parallelism.
377
+ * When maxInFlightBatches > 1, processes multiple batches concurrently.
378
+ */
379
+ embedBatch(texts: string[]): Promise<Float32Array[]>;
380
+ /**
381
+ * Process batches sequentially (original behavior).
382
+ */
383
+ private embedBatchesSequential;
384
+ /**
385
+ * Process batches with controlled concurrency.
386
+ */
387
+ private embedBatchesConcurrent;
388
+ /**
389
+ * Process a single batch and return embeddings.
390
+ */
391
+ private processSingleBatch;
360
392
  /**
361
393
  * Get cached embedding dimensions. Throws if embed() hasn't been called yet.
362
394
  * Use ensureDimensions() if you need to guarantee dimensions are available.
363
395
  */
364
396
  getDimensions(): number;
397
+ /**
398
+ * Check if the embedding pipeline is initialized.
399
+ */
400
+ isInitialized(): boolean;
401
+ /**
402
+ * Check if this engine has been disposed.
403
+ */
404
+ isDisposed(): boolean;
405
+ /**
406
+ * Reset the engine to uninitialized state, allowing reuse after disposal.
407
+ * If currently initialized, disposes the pipeline first.
408
+ */
409
+ reset(): Promise<void>;
365
410
  /**
366
411
  * Ensure dimensions are available, initializing the model if needed.
367
412
  * Returns the embedding dimensions for the current model.
@@ -395,7 +440,7 @@ interface DocumentMetadata {
395
440
  interface Document {
396
441
  readonly id: DocumentId;
397
442
  readonly content: string;
398
- readonly vector: readonly number[];
443
+ readonly vector: Float32Array;
399
444
  readonly metadata: DocumentMetadata;
400
445
  }
401
446
 
@@ -404,7 +449,18 @@ declare class LanceStore {
404
449
  private readonly tables;
405
450
  private readonly dataDir;
406
451
  private _dimensions;
452
+ private embeddingFunction;
407
453
  constructor(dataDir: string);
454
+ /**
455
+ * Set the embedding function for auto-embedding queries.
456
+ * Must be called before initialize() for new tables.
457
+ * The embedding function is initialized and its dimensions are used for schema creation.
458
+ */
459
+ setEmbeddingFunction(config: EmbeddingConfig): Promise<void>;
460
+ /**
461
+ * Check if embedding function is available for auto-embedding queries.
462
+ */
463
+ hasEmbeddingFunction(): boolean;
408
464
  /**
409
465
  * Set the embedding dimensions. Must be called before initialize().
410
466
  * This allows dimensions to be derived from the embedding model at runtime.
@@ -415,7 +471,18 @@ declare class LanceStore {
415
471
  addDocuments(storeId: StoreId, documents: Document[]): Promise<void>;
416
472
  deleteDocuments(storeId: StoreId, documentIds: DocumentId[]): Promise<void>;
417
473
  clearAllDocuments(storeId: StoreId): Promise<void>;
418
- search(storeId: StoreId, vector: number[], limit: number, _threshold?: number): Promise<Array<{
474
+ search(storeId: StoreId, vector: Float32Array, limit: number, _threshold?: number): Promise<Array<{
475
+ id: DocumentId;
476
+ content: string;
477
+ score: number;
478
+ metadata: DocumentMetadata;
479
+ }>>;
480
+ /**
481
+ * Search using a text query with automatic embedding.
482
+ * Requires setEmbeddingFunction() to have been called.
483
+ * Uses the embedding function to compute query embeddings consistently with document embeddings.
484
+ */
485
+ searchText(storeId: StoreId, query: string, limit: number): Promise<Array<{
419
486
  id: DocumentId;
420
487
  content: string;
421
488
  score: number;
@@ -467,6 +534,16 @@ interface BaseStore {
467
534
  readonly status?: StoreStatus | undefined;
468
535
  readonly createdAt: Date;
469
536
  readonly updatedAt: Date;
537
+ /**
538
+ * Schema version for detecting incompatible stores.
539
+ * Stores with schemaVersion < CURRENT_SCHEMA_VERSION require reindex.
540
+ */
541
+ readonly schemaVersion: number;
542
+ /**
543
+ * Embedding model ID used to index this store.
544
+ * Used for provenance tracking and mismatch detection.
545
+ */
546
+ readonly modelId: string;
470
547
  }
471
548
  interface FileStore extends BaseStore {
472
549
  readonly type: 'file';
@@ -638,13 +715,12 @@ interface SearchResponse {
638
715
 
639
716
  declare class SearchService {
640
717
  private readonly lanceStore;
641
- private readonly embeddingEngine;
642
718
  private readonly codeUnitService;
643
719
  private readonly codeGraphService;
644
720
  private readonly graphCache;
645
721
  private readonly searchConfig;
646
722
  private readonly unsubscribeCacheInvalidation;
647
- constructor(lanceStore: LanceStore, embeddingEngine: EmbeddingEngine, codeGraphService?: CodeGraphService, searchConfig?: SearchConfig);
723
+ constructor(lanceStore: LanceStore, codeGraphService?: CodeGraphService, searchConfig?: SearchConfig);
648
724
  /**
649
725
  * Clean up resources (unsubscribe from events).
650
726
  * Call this when destroying the service.
@@ -683,6 +759,8 @@ declare class SearchService {
683
759
  /**
684
760
  * Fetch raw vector search results without normalization.
685
761
  * Returns results with raw cosine similarity scores [0-1].
762
+ * Uses LanceDB's embedding function for query embedding,
763
+ * ensuring consistent query/document embedding through a single code path.
686
764
  */
687
765
  private vectorSearchRaw;
688
766
  private ftsSearch;
@@ -930,6 +1008,8 @@ interface StoreServiceOptions {
930
1008
  gitignoreService?: GitignoreService;
931
1009
  /** Optional project root for resolving relative paths */
932
1010
  projectRoot?: string;
1011
+ /** Embedding model ID for provenance tracking (required for new stores) */
1012
+ embeddingModelId: string;
933
1013
  }
934
1014
  interface OperationOptions {
935
1015
  /** Skip syncing to store definitions (used by stores:sync command) */
@@ -940,8 +1020,14 @@ declare class StoreService {
940
1020
  private readonly definitionService;
941
1021
  private readonly gitignoreService;
942
1022
  private readonly projectRoot;
1023
+ private readonly embeddingModelId;
943
1024
  private registry;
944
- constructor(dataDir: string, options?: StoreServiceOptions);
1025
+ constructor(dataDir: string, options: StoreServiceOptions);
1026
+ /**
1027
+ * Get the current embedding model ID used for new stores.
1028
+ * Used by model compatibility validation.
1029
+ */
1030
+ getCurrentModelId(): string;
945
1031
  initialize(): Promise<void>;
946
1032
  /**
947
1033
  * Convert a Store and CreateStoreInput to a StoreDefinition for persistence.
@@ -1,10 +1,10 @@
1
1
  import {
2
2
  createMCPServer,
3
3
  runMCPServer
4
- } from "../chunk-6BC5OG4M.js";
5
- import "../chunk-JPJI3VMA.js";
4
+ } from "../chunk-BZQ7KWEE.js";
5
+ import "../chunk-ZR23KJPJ.js";
6
6
  import "../chunk-CLIMKLTW.js";
7
- import "../chunk-HXBIIMYL.js";
7
+ import "../chunk-H25AEF47.js";
8
8
  import "../chunk-DGUM43GV.js";
9
9
  export {
10
10
  createMCPServer,
@@ -1,8 +1,8 @@
1
1
  import {
2
2
  WatchService
3
- } from "./chunk-HXBIIMYL.js";
3
+ } from "./chunk-H25AEF47.js";
4
4
  import "./chunk-DGUM43GV.js";
5
5
  export {
6
6
  WatchService
7
7
  };
8
- //# sourceMappingURL=watch.service-NXRWLJG6.js.map
8
+ //# sourceMappingURL=watch.service-THP6X5ZZ.js.map
@@ -1,19 +1,19 @@
1
1
  #!/usr/bin/env node
2
2
  import {
3
3
  IntelligentCrawler
4
- } from "../chunk-TWX7MN5L.js";
4
+ } from "../chunk-VNHZ534Q.js";
5
5
  import {
6
6
  JobService,
7
7
  createLogger,
8
8
  createServices,
9
9
  destroyServices,
10
10
  shutdownLogger
11
- } from "../chunk-JPJI3VMA.js";
11
+ } from "../chunk-ZR23KJPJ.js";
12
12
  import {
13
13
  createDocumentId,
14
14
  createStoreId
15
15
  } from "../chunk-CLIMKLTW.js";
16
- import "../chunk-HXBIIMYL.js";
16
+ import "../chunk-H25AEF47.js";
17
17
  import "../chunk-DGUM43GV.js";
18
18
 
19
19
  // src/workers/background-worker-cli.ts
@@ -222,7 +222,7 @@ var BackgroundWorker = class {
222
222
  throw new Error("Job cancelled by user");
223
223
  }
224
224
  const contentToEmbed = result.extracted ?? result.markdown;
225
- const vector = await this.embeddingEngine.embed(contentToEmbed);
225
+ const vector = await this.embeddingEngine.embedDocument(contentToEmbed);
226
226
  docs.push({
227
227
  id: createDocumentId(`${store.id}-${createHash("md5").update(result.url).digest("hex")}`),
228
228
  content: contentToEmbed,
@@ -1 +1 @@
1
- {"version":3,"sources":["../../src/workers/background-worker-cli.ts","../../src/workers/background-worker.ts","../../src/workers/pid-file.ts"],"sourcesContent":["#!/usr/bin/env node\nimport { platform } from 'os';\nimport { BackgroundWorker } from './background-worker.js';\nimport { writePidFile, deletePidFile, buildPidFilePath } from './pid-file.js';\nimport { createLogger, shutdownLogger } from '../logging/index.js';\nimport { createServices, destroyServices } from '../services/index.js';\nimport { JobService } from '../services/job.service.js';\n\n/**\n * Force exit the process to avoid ONNX runtime mutex crash on macOS.\n *\n * On macOS, the ONNX runtime (used by transformers.js for embeddings) has a known\n * bug where static mutex cleanup fails during process exit, causing a crash with:\n * \"mutex lock failed: Invalid argument\"\n *\n * This doesn't affect job completion - all work is done and persisted before exit.\n * Using SIGKILL bypasses the problematic cleanup code.\n *\n * See: https://github.com/microsoft/onnxruntime/issues/24579\n */\nfunction forceExitOnMacOS(exitCode: number): void {\n if (platform() === 'darwin') {\n // Give time for any pending I/O to flush\n setTimeout(() => {\n process.kill(process.pid, 'SIGKILL');\n }, 100);\n } else {\n process.exit(exitCode);\n }\n}\n\nconst logger = createLogger('background-worker-cli');\n\n/**\n * Background worker CLI entry point\n *\n * Usage: background-worker-cli <job-id>\n *\n * This process runs detached from the parent and executes a single job.\n */\n\nasync function main(): Promise<void> {\n const jobId = process.argv[2];\n const dataDir = process.env['BLUERA_DATA_DIR'];\n\n if (jobId === undefined || jobId === '') {\n logger.error('Job ID required. Usage: background-worker-cli <job-id>');\n await shutdownLogger();\n process.exit(1);\n }\n\n // Initialize services\n const jobService = new JobService(dataDir);\n const services = await createServices(undefined, dataDir);\n\n // Write PID file for job cancellation - CRITICAL: must succeed or job cannot be cancelled\n const pidFile = buildPidFilePath(\n jobService['jobsDir'], // Access private field for PID path\n jobId\n );\n\n try {\n writePidFile(pidFile, process.pid);\n } catch (error) {\n // CRITICAL: Cannot proceed without PID file - job would be uncancellable\n logger.error(\n { error: error instanceof Error ? error.message : String(error) },\n 'Failed to write PID file'\n );\n await shutdownLogger();\n process.exit(1);\n }\n\n // Handle SIGTERM for graceful shutdown\n process.on('SIGTERM', () => {\n logger.info({ jobId }, 'Received SIGTERM, cancelling job');\n jobService.updateJob(jobId, {\n status: 'cancelled',\n message: 'Job cancelled by user',\n });\n\n // Clean up PID file (best-effort - don't block shutdown)\n const deleteResult = deletePidFile(pidFile, 'sigterm');\n if (!deleteResult.success && deleteResult.error !== undefined) {\n logger.warn(\n { jobId, error: deleteResult.error.message },\n 'Could not remove PID file during SIGTERM'\n );\n }\n\n // Flush logs before exit (best-effort, don't await in signal handler)\n void shutdownLogger().finally(() => process.exit(0));\n });\n\n // Load config and create worker\n const appConfig = await services.config.load();\n const worker = new BackgroundWorker(\n jobService,\n services.store,\n services.index,\n services.lance,\n services.embeddings,\n appConfig.crawl\n );\n\n try {\n await worker.executeJob(jobId);\n\n // Clean up PID file on success (best-effort - don't change exit code)\n const successCleanup = deletePidFile(pidFile, 'success');\n if (!successCleanup.success && successCleanup.error !== undefined) {\n logger.warn(\n { jobId, error: successCleanup.error.message },\n 'Could not remove PID file after success'\n );\n }\n\n logger.info({ jobId }, 'Job completed successfully');\n await destroyServices(services);\n await shutdownLogger();\n forceExitOnMacOS(0);\n } catch (error) {\n // Job service already updated with failure status in BackgroundWorker\n logger.error(\n { jobId, error: error instanceof Error ? error.message : String(error) },\n 'Job failed'\n );\n\n // Clean up PID file on failure (best-effort - exit code reflects job failure)\n const failureCleanup = deletePidFile(pidFile, 'failure');\n if (!failureCleanup.success && failureCleanup.error !== undefined) {\n logger.warn(\n { jobId, error: failureCleanup.error.message },\n 'Could not remove PID file after failure'\n );\n }\n\n await destroyServices(services);\n await shutdownLogger();\n forceExitOnMacOS(1);\n }\n}\n\nmain().catch(async (error: unknown) => {\n logger.error(\n { error: error instanceof Error ? error.message : String(error) },\n 'Fatal error in background worker'\n );\n await shutdownLogger();\n forceExitOnMacOS(1);\n});\n","import { createHash } from 'node:crypto';\nimport {\n IntelligentCrawler,\n type CrawlConfig,\n type CrawlProgress,\n} from '../crawl/intelligent-crawler.js';\nimport { createLogger } from '../logging/index.js';\nimport { IndexService } from '../services/index.service.js';\nimport { JobService } from '../services/job.service.js';\nimport { StoreService } from '../services/store.service.js';\nimport { createStoreId, createDocumentId } from '../types/brands.js';\nimport type { EmbeddingEngine } from '../db/embeddings.js';\nimport type { LanceStore } from '../db/lance.js';\nimport type { Document } from '../types/document.js';\nimport type { Job } from '../types/job.js';\n\nconst logger = createLogger('background-worker');\n\n/**\n * Calculate index progress as a percentage, handling division by zero.\n * @param current - Current number of items processed\n * @param total - Total number of items (may be 0)\n * @param scale - Scale factor for progress (default 100 for 0-100%)\n * @returns Progress value, or 0 if total is 0\n */\nexport function calculateIndexProgress(\n current: number,\n total: number,\n scale: number = 100\n): number {\n if (total === 0) return 0;\n return (current / total) * scale;\n}\n\nexport class BackgroundWorker {\n constructor(\n private readonly jobService: JobService,\n private readonly storeService: StoreService,\n private readonly indexService: IndexService,\n private readonly lanceStore: LanceStore,\n private readonly embeddingEngine: EmbeddingEngine,\n private readonly crawlConfig?: CrawlConfig\n ) {}\n\n /**\n * Execute a job based on its type\n */\n async executeJob(jobId: string): Promise<void> {\n const job = this.jobService.getJob(jobId);\n\n if (!job) {\n throw new Error(`Job ${jobId} not found`);\n }\n\n try {\n logger.info({ jobId, type: job.type }, 'Starting job execution');\n\n // Update to running status\n this.jobService.updateJob(jobId, {\n status: 'running',\n message: `Starting ${job.type} operation...`,\n progress: 0,\n details: { startedAt: new Date().toISOString() },\n });\n\n // Execute based on job type\n switch (job.type) {\n case 'clone':\n await this.executeCloneJob(job);\n break;\n case 'index':\n await this.executeIndexJob(job);\n break;\n case 'crawl':\n await this.executeCrawlJob(job);\n break;\n default:\n throw new Error(`Unknown job type: ${String(job.type)}`);\n }\n\n // Mark as completed\n this.jobService.updateJob(jobId, {\n status: 'completed',\n progress: 100,\n message: `${job.type} operation completed successfully`,\n details: { completedAt: new Date().toISOString() },\n });\n } catch (error) {\n logger.error(\n { jobId, error: error instanceof Error ? error.message : String(error) },\n 'Job failed'\n );\n\n // Mark as failed\n const errorDetails: Record<string, unknown> = {\n completedAt: new Date().toISOString(),\n };\n if (error instanceof Error && error.stack !== undefined) {\n errorDetails['error'] = error.stack;\n } else {\n errorDetails['error'] = String(error);\n }\n this.jobService.updateJob(jobId, {\n status: 'failed',\n message: error instanceof Error ? error.message : 'Unknown error',\n details: errorDetails,\n });\n throw error;\n }\n }\n\n /**\n * Execute a clone job (git clone + initial indexing)\n */\n private async executeCloneJob(job: Job): Promise<void> {\n const { storeId } = job.details;\n\n if (storeId === undefined || typeof storeId !== 'string') {\n throw new Error('Store ID required for clone job');\n }\n\n // Get the store\n const store = await this.storeService.get(createStoreId(storeId));\n if (!store) {\n throw new Error(`Store ${storeId} not found`);\n }\n\n // Clone is already done by the time the job is created\n // (happens in StoreService.create), so we just need to index\n\n // Update progress - cloning considered done (30%)\n this.jobService.updateJob(job.id, {\n status: 'running',\n message: 'Repository cloned, starting indexing...',\n progress: 30,\n });\n\n // Initialize LanceStore with dimensions before indexing\n this.lanceStore.setDimensions(await this.embeddingEngine.ensureDimensions());\n await this.lanceStore.initialize(store.id);\n\n // Index the repository with progress updates\n const result = await this.indexService.indexStore(\n store,\n (event: { type: string; current: number; total: number; message: string }) => {\n // Check if job was cancelled\n const currentJob = this.jobService.getJob(job.id);\n if (currentJob?.status === 'cancelled') {\n throw new Error('Job cancelled by user');\n }\n\n // Indexing is 70% of total progress (30-100%)\n const indexProgress = calculateIndexProgress(event.current, event.total, 70);\n const totalProgress = 30 + indexProgress;\n\n this.jobService.updateJob(job.id, {\n message: `Indexed ${String(event.current)}/${String(event.total)} files`,\n progress: Math.min(99, totalProgress), // Cap at 99 until fully complete\n details: {\n filesProcessed: event.current,\n totalFiles: event.total,\n },\n });\n }\n );\n\n if (!result.success) {\n throw result.error;\n }\n }\n\n /**\n * Execute an index job (re-indexing existing store)\n */\n private async executeIndexJob(job: Job): Promise<void> {\n const { storeId } = job.details;\n\n if (storeId === undefined || typeof storeId !== 'string') {\n throw new Error('Store ID required for index job');\n }\n\n // Get the store\n const store = await this.storeService.getByIdOrName(createStoreId(storeId));\n if (!store) {\n throw new Error(`Store ${storeId} not found`);\n }\n\n // Initialize LanceStore with dimensions before indexing\n this.lanceStore.setDimensions(await this.embeddingEngine.ensureDimensions());\n await this.lanceStore.initialize(store.id);\n\n // Index with progress updates\n const result = await this.indexService.indexStore(\n store,\n (event: { type: string; current: number; total: number; message: string }) => {\n // Check if job was cancelled\n const currentJob = this.jobService.getJob(job.id);\n if (currentJob?.status === 'cancelled') {\n throw new Error('Job cancelled by user');\n }\n\n const progress = calculateIndexProgress(event.current, event.total);\n\n this.jobService.updateJob(job.id, {\n message: `Indexed ${String(event.current)}/${String(event.total)} files`,\n progress: Math.min(99, progress), // Cap at 99 until fully complete\n details: {\n filesProcessed: event.current,\n totalFiles: event.total,\n },\n });\n }\n );\n\n if (!result.success) {\n throw result.error;\n }\n }\n\n /**\n * Execute a crawl job (web crawling + indexing)\n */\n private async executeCrawlJob(job: Job): Promise<void> {\n const { storeId, url, crawlInstruction, extractInstruction, maxPages, simple, useHeadless } =\n job.details;\n\n if (storeId === undefined || typeof storeId !== 'string') {\n throw new Error('Store ID required for crawl job');\n }\n if (url === undefined || typeof url !== 'string') {\n throw new Error('URL required for crawl job');\n }\n\n // Get the store\n const store = await this.storeService.get(createStoreId(storeId));\n if (store?.type !== 'web') {\n throw new Error(`Web store ${storeId} not found`);\n }\n\n const resolvedMaxPages = typeof maxPages === 'number' ? maxPages : 50;\n const crawler = new IntelligentCrawler(this.crawlConfig);\n\n // Listen for progress events\n crawler.on('progress', (progress: CrawlProgress) => {\n // Check if job was cancelled - just return early, for-await loop will throw and finally will cleanup\n const currentJob = this.jobService.getJob(job.id);\n if (currentJob?.status === 'cancelled') {\n return;\n }\n\n // Crawling is 80% of total progress (0-80%)\n const crawlProgress = (progress.pagesVisited / resolvedMaxPages) * 80;\n\n this.jobService.updateJob(job.id, {\n message:\n progress.message ??\n `Crawling page ${String(progress.pagesVisited)}/${String(resolvedMaxPages)}`,\n progress: Math.min(80, crawlProgress),\n details: { pagesCrawled: progress.pagesVisited },\n });\n });\n\n try {\n this.lanceStore.setDimensions(await this.embeddingEngine.ensureDimensions());\n await this.lanceStore.initialize(store.id);\n const docs: Document[] = [];\n\n // Build crawl options, only including defined values\n const crawlOptions: {\n maxPages: number;\n simple: boolean;\n useHeadless: boolean;\n crawlInstruction?: string;\n extractInstruction?: string;\n } = {\n maxPages: resolvedMaxPages,\n simple: simple ?? false,\n useHeadless: useHeadless ?? true, // Default to headless for reliability\n };\n if (crawlInstruction !== undefined) {\n crawlOptions.crawlInstruction = crawlInstruction;\n }\n if (extractInstruction !== undefined) {\n crawlOptions.extractInstruction = extractInstruction;\n }\n\n // Crawl pages using IntelligentCrawler\n for await (const result of crawler.crawl(url, crawlOptions)) {\n // Check cancellation between pages\n const currentJob = this.jobService.getJob(job.id);\n if (currentJob?.status === 'cancelled') {\n throw new Error('Job cancelled by user');\n }\n\n // Embed and index the content (use extracted if available, otherwise markdown)\n const contentToEmbed = result.extracted ?? result.markdown;\n const vector = await this.embeddingEngine.embed(contentToEmbed);\n\n docs.push({\n id: createDocumentId(`${store.id}-${createHash('md5').update(result.url).digest('hex')}`),\n content: contentToEmbed,\n vector,\n metadata: {\n type: 'web',\n storeId: store.id,\n url: result.url,\n title: result.title,\n extracted: result.extracted !== undefined,\n depth: result.depth,\n indexedAt: new Date().toISOString(),\n },\n });\n }\n\n // Index all documents (remaining 20%)\n if (docs.length > 0) {\n this.jobService.updateJob(job.id, {\n message: 'Indexing crawled documents...',\n progress: 85,\n });\n\n // Clear existing documents to prevent duplicates on re-crawl\n await this.lanceStore.clearAllDocuments(store.id);\n await this.lanceStore.addDocuments(store.id, docs);\n // Create FTS index for full-text search\n await this.lanceStore.createFtsIndex(store.id);\n }\n\n this.jobService.updateJob(job.id, {\n message: `Crawled and indexed ${String(docs.length)} pages`,\n progress: 100,\n details: { pagesCrawled: docs.length },\n });\n } finally {\n await crawler.stop();\n }\n }\n}\n","import fs from 'fs';\nimport path from 'path';\n\n/**\n * Result of a PID file delete operation.\n * Delete operations are best-effort and should not throw.\n */\nexport interface PidFileResult {\n success: boolean;\n error?: Error;\n}\n\n/**\n * Context for PID file deletion - indicates when the delete is happening.\n * Used for logging/debugging purposes.\n */\nexport type PidFileDeleteContext = 'sigterm' | 'success' | 'failure';\n\n/**\n * Write PID file - CRITICAL operation that must succeed.\n *\n * If the PID file cannot be written, the job cannot be cancelled through\n * the job management system. This is a critical failure and the job\n * should not proceed.\n *\n * @param pidFile - Absolute path to the PID file\n * @param pid - Process ID to write\n * @throws Error if PID file cannot be written\n */\nexport function writePidFile(pidFile: string, pid: number): void {\n try {\n fs.writeFileSync(pidFile, pid.toString(), 'utf-8');\n } catch (error) {\n const message = error instanceof Error ? error.message : String(error);\n throw new Error(\n `CRITICAL: Failed to write PID file ${pidFile}. ` +\n `Job cannot be cancelled without PID file. ` +\n `Original error: ${message}`\n );\n }\n}\n\n/**\n * Delete PID file - best-effort cleanup during shutdown.\n *\n * This operation should NEVER throw. During process shutdown (SIGTERM,\n * job success, job failure), failing to delete a PID file should not\n * prevent the process from exiting cleanly.\n *\n * Stale PID files are cleaned up by JobService.cleanupOldJobs().\n *\n * @param pidFile - Absolute path to the PID file\n * @param _context - Context indicating when the delete is happening (for future logging)\n * @returns Result indicating success or failure with error details\n */\nexport function deletePidFile(pidFile: string, _context: PidFileDeleteContext): PidFileResult {\n try {\n fs.unlinkSync(pidFile);\n return { success: true };\n } catch (error) {\n // ENOENT = file doesn't exist - that's success (nothing to delete)\n if (error instanceof Error && 'code' in error && error.code === 'ENOENT') {\n return { success: true };\n }\n // Any other error = failure (permission denied, etc.)\n return {\n success: false,\n error: error instanceof Error ? error : new Error(String(error)),\n };\n }\n}\n\n/**\n * Build the path to a PID file for a given job.\n *\n * @param jobsDir - Directory where job files are stored\n * @param jobId - Job identifier\n * @returns Absolute path to the PID file\n */\nexport function buildPidFilePath(jobsDir: string, jobId: string): string {\n return path.join(jobsDir, `${jobId}.pid`);\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;;AACA,SAAS,gBAAgB;;;ACDzB,SAAS,kBAAkB;AAgB3B,IAAM,SAAS,aAAa,mBAAmB;AASxC,SAAS,uBACd,SACA,OACA,QAAgB,KACR;AACR,MAAI,UAAU,EAAG,QAAO;AACxB,SAAQ,UAAU,QAAS;AAC7B;AAEO,IAAM,mBAAN,MAAuB;AAAA,EAC5B,YACmB,YACA,cACA,cACA,YACA,iBACA,aACjB;AANiB;AACA;AACA;AACA;AACA;AACA;AAAA,EAChB;AAAA;AAAA;AAAA;AAAA,EAKH,MAAM,WAAW,OAA8B;AAC7C,UAAM,MAAM,KAAK,WAAW,OAAO,KAAK;AAExC,QAAI,CAAC,KAAK;AACR,YAAM,IAAI,MAAM,OAAO,KAAK,YAAY;AAAA,IAC1C;AAEA,QAAI;AACF,aAAO,KAAK,EAAE,OAAO,MAAM,IAAI,KAAK,GAAG,wBAAwB;AAG/D,WAAK,WAAW,UAAU,OAAO;AAAA,QAC/B,QAAQ;AAAA,QACR,SAAS,YAAY,IAAI,IAAI;AAAA,QAC7B,UAAU;AAAA,QACV,SAAS,EAAE,YAAW,oBAAI,KAAK,GAAE,YAAY,EAAE;AAAA,MACjD,CAAC;AAGD,cAAQ,IAAI,MAAM;AAAA,QAChB,KAAK;AACH,gBAAM,KAAK,gBAAgB,GAAG;AAC9B;AAAA,QACF,KAAK;AACH,gBAAM,KAAK,gBAAgB,GAAG;AAC9B;AAAA,QACF,KAAK;AACH,gBAAM,KAAK,gBAAgB,GAAG;AAC9B;AAAA,QACF;AACE,gBAAM,IAAI,MAAM,qBAAqB,OAAO,IAAI,IAAI,CAAC,EAAE;AAAA,MAC3D;AAGA,WAAK,WAAW,UAAU,OAAO;AAAA,QAC/B,QAAQ;AAAA,QACR,UAAU;AAAA,QACV,SAAS,GAAG,IAAI,IAAI;AAAA,QACpB,SAAS,EAAE,cAAa,oBAAI,KAAK,GAAE,YAAY,EAAE;AAAA,MACnD,CAAC;AAAA,IACH,SAAS,OAAO;AACd,aAAO;AAAA,QACL,EAAE,OAAO,OAAO,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK,EAAE;AAAA,QACvE;AAAA,MACF;AAGA,YAAM,eAAwC;AAAA,QAC5C,cAAa,oBAAI,KAAK,GAAE,YAAY;AAAA,MACtC;AACA,UAAI,iBAAiB,SAAS,MAAM,UAAU,QAAW;AACvD,qBAAa,OAAO,IAAI,MAAM;AAAA,MAChC,OAAO;AACL,qBAAa,OAAO,IAAI,OAAO,KAAK;AAAA,MACtC;AACA,WAAK,WAAW,UAAU,OAAO;AAAA,QAC/B,QAAQ;AAAA,QACR,SAAS,iBAAiB,QAAQ,MAAM,UAAU;AAAA,QAClD,SAAS;AAAA,MACX,CAAC;AACD,YAAM;AAAA,IACR;AAAA,EACF;AAAA;AAAA;AAAA;AAAA,EAKA,MAAc,gBAAgB,KAAyB;AACrD,UAAM,EAAE,QAAQ,IAAI,IAAI;AAExB,QAAI,YAAY,UAAa,OAAO,YAAY,UAAU;AACxD,YAAM,IAAI,MAAM,iCAAiC;AAAA,IACnD;AAGA,UAAM,QAAQ,MAAM,KAAK,aAAa,IAAI,cAAc,OAAO,CAAC;AAChE,QAAI,CAAC,OAAO;AACV,YAAM,IAAI,MAAM,SAAS,OAAO,YAAY;AAAA,IAC9C;AAMA,SAAK,WAAW,UAAU,IAAI,IAAI;AAAA,MAChC,QAAQ;AAAA,MACR,SAAS;AAAA,MACT,UAAU;AAAA,IACZ,CAAC;AAGD,SAAK,WAAW,cAAc,MAAM,KAAK,gBAAgB,iBAAiB,CAAC;AAC3E,UAAM,KAAK,WAAW,WAAW,MAAM,EAAE;AAGzC,UAAM,SAAS,MAAM,KAAK,aAAa;AAAA,MACrC;AAAA,MACA,CAAC,UAA6E;AAE5E,cAAM,aAAa,KAAK,WAAW,OAAO,IAAI,EAAE;AAChD,YAAI,YAAY,WAAW,aAAa;AACtC,gBAAM,IAAI,MAAM,uBAAuB;AAAA,QACzC;AAGA,cAAM,gBAAgB,uBAAuB,MAAM,SAAS,MAAM,OAAO,EAAE;AAC3E,cAAM,gBAAgB,KAAK;AAE3B,aAAK,WAAW,UAAU,IAAI,IAAI;AAAA,UAChC,SAAS,WAAW,OAAO,MAAM,OAAO,CAAC,IAAI,OAAO,MAAM,KAAK,CAAC;AAAA,UAChE,UAAU,KAAK,IAAI,IAAI,aAAa;AAAA;AAAA,UACpC,SAAS;AAAA,YACP,gBAAgB,MAAM;AAAA,YACtB,YAAY,MAAM;AAAA,UACpB;AAAA,QACF,CAAC;AAAA,MACH;AAAA,IACF;AAEA,QAAI,CAAC,OAAO,SAAS;AACnB,YAAM,OAAO;AAAA,IACf;AAAA,EACF;AAAA;AAAA;AAAA;AAAA,EAKA,MAAc,gBAAgB,KAAyB;AACrD,UAAM,EAAE,QAAQ,IAAI,IAAI;AAExB,QAAI,YAAY,UAAa,OAAO,YAAY,UAAU;AACxD,YAAM,IAAI,MAAM,iCAAiC;AAAA,IACnD;AAGA,UAAM,QAAQ,MAAM,KAAK,aAAa,cAAc,cAAc,OAAO,CAAC;AAC1E,QAAI,CAAC,OAAO;AACV,YAAM,IAAI,MAAM,SAAS,OAAO,YAAY;AAAA,IAC9C;AAGA,SAAK,WAAW,cAAc,MAAM,KAAK,gBAAgB,iBAAiB,CAAC;AAC3E,UAAM,KAAK,WAAW,WAAW,MAAM,EAAE;AAGzC,UAAM,SAAS,MAAM,KAAK,aAAa;AAAA,MACrC;AAAA,MACA,CAAC,UAA6E;AAE5E,cAAM,aAAa,KAAK,WAAW,OAAO,IAAI,EAAE;AAChD,YAAI,YAAY,WAAW,aAAa;AACtC,gBAAM,IAAI,MAAM,uBAAuB;AAAA,QACzC;AAEA,cAAM,WAAW,uBAAuB,MAAM,SAAS,MAAM,KAAK;AAElE,aAAK,WAAW,UAAU,IAAI,IAAI;AAAA,UAChC,SAAS,WAAW,OAAO,MAAM,OAAO,CAAC,IAAI,OAAO,MAAM,KAAK,CAAC;AAAA,UAChE,UAAU,KAAK,IAAI,IAAI,QAAQ;AAAA;AAAA,UAC/B,SAAS;AAAA,YACP,gBAAgB,MAAM;AAAA,YACtB,YAAY,MAAM;AAAA,UACpB;AAAA,QACF,CAAC;AAAA,MACH;AAAA,IACF;AAEA,QAAI,CAAC,OAAO,SAAS;AACnB,YAAM,OAAO;AAAA,IACf;AAAA,EACF;AAAA;AAAA;AAAA;AAAA,EAKA,MAAc,gBAAgB,KAAyB;AACrD,UAAM,EAAE,SAAS,KAAK,kBAAkB,oBAAoB,UAAU,QAAQ,YAAY,IACxF,IAAI;AAEN,QAAI,YAAY,UAAa,OAAO,YAAY,UAAU;AACxD,YAAM,IAAI,MAAM,iCAAiC;AAAA,IACnD;AACA,QAAI,QAAQ,UAAa,OAAO,QAAQ,UAAU;AAChD,YAAM,IAAI,MAAM,4BAA4B;AAAA,IAC9C;AAGA,UAAM,QAAQ,MAAM,KAAK,aAAa,IAAI,cAAc,OAAO,CAAC;AAChE,QAAI,OAAO,SAAS,OAAO;AACzB,YAAM,IAAI,MAAM,aAAa,OAAO,YAAY;AAAA,IAClD;AAEA,UAAM,mBAAmB,OAAO,aAAa,WAAW,WAAW;AACnE,UAAM,UAAU,IAAI,mBAAmB,KAAK,WAAW;AAGvD,YAAQ,GAAG,YAAY,CAAC,aAA4B;AAElD,YAAM,aAAa,KAAK,WAAW,OAAO,IAAI,EAAE;AAChD,UAAI,YAAY,WAAW,aAAa;AACtC;AAAA,MACF;AAGA,YAAM,gBAAiB,SAAS,eAAe,mBAAoB;AAEnE,WAAK,WAAW,UAAU,IAAI,IAAI;AAAA,QAChC,SACE,SAAS,WACT,iBAAiB,OAAO,SAAS,YAAY,CAAC,IAAI,OAAO,gBAAgB,CAAC;AAAA,QAC5E,UAAU,KAAK,IAAI,IAAI,aAAa;AAAA,QACpC,SAAS,EAAE,cAAc,SAAS,aAAa;AAAA,MACjD,CAAC;AAAA,IACH,CAAC;AAED,QAAI;AACF,WAAK,WAAW,cAAc,MAAM,KAAK,gBAAgB,iBAAiB,CAAC;AAC3E,YAAM,KAAK,WAAW,WAAW,MAAM,EAAE;AACzC,YAAM,OAAmB,CAAC;AAG1B,YAAM,eAMF;AAAA,QACF,UAAU;AAAA,QACV,QAAQ,UAAU;AAAA,QAClB,aAAa,eAAe;AAAA;AAAA,MAC9B;AACA,UAAI,qBAAqB,QAAW;AAClC,qBAAa,mBAAmB;AAAA,MAClC;AACA,UAAI,uBAAuB,QAAW;AACpC,qBAAa,qBAAqB;AAAA,MACpC;AAGA,uBAAiB,UAAU,QAAQ,MAAM,KAAK,YAAY,GAAG;AAE3D,cAAM,aAAa,KAAK,WAAW,OAAO,IAAI,EAAE;AAChD,YAAI,YAAY,WAAW,aAAa;AACtC,gBAAM,IAAI,MAAM,uBAAuB;AAAA,QACzC;AAGA,cAAM,iBAAiB,OAAO,aAAa,OAAO;AAClD,cAAM,SAAS,MAAM,KAAK,gBAAgB,MAAM,cAAc;AAE9D,aAAK,KAAK;AAAA,UACR,IAAI,iBAAiB,GAAG,MAAM,EAAE,IAAI,WAAW,KAAK,EAAE,OAAO,OAAO,GAAG,EAAE,OAAO,KAAK,CAAC,EAAE;AAAA,UACxF,SAAS;AAAA,UACT;AAAA,UACA,UAAU;AAAA,YACR,MAAM;AAAA,YACN,SAAS,MAAM;AAAA,YACf,KAAK,OAAO;AAAA,YACZ,OAAO,OAAO;AAAA,YACd,WAAW,OAAO,cAAc;AAAA,YAChC,OAAO,OAAO;AAAA,YACd,YAAW,oBAAI,KAAK,GAAE,YAAY;AAAA,UACpC;AAAA,QACF,CAAC;AAAA,MACH;AAGA,UAAI,KAAK,SAAS,GAAG;AACnB,aAAK,WAAW,UAAU,IAAI,IAAI;AAAA,UAChC,SAAS;AAAA,UACT,UAAU;AAAA,QACZ,CAAC;AAGD,cAAM,KAAK,WAAW,kBAAkB,MAAM,EAAE;AAChD,cAAM,KAAK,WAAW,aAAa,MAAM,IAAI,IAAI;AAEjD,cAAM,KAAK,WAAW,eAAe,MAAM,EAAE;AAAA,MAC/C;AAEA,WAAK,WAAW,UAAU,IAAI,IAAI;AAAA,QAChC,SAAS,uBAAuB,OAAO,KAAK,MAAM,CAAC;AAAA,QACnD,UAAU;AAAA,QACV,SAAS,EAAE,cAAc,KAAK,OAAO;AAAA,MACvC,CAAC;AAAA,IACH,UAAE;AACA,YAAM,QAAQ,KAAK;AAAA,IACrB;AAAA,EACF;AACF;;;ACjVA,OAAO,QAAQ;AACf,OAAO,UAAU;AA4BV,SAAS,aAAa,SAAiB,KAAmB;AAC/D,MAAI;AACF,OAAG,cAAc,SAAS,IAAI,SAAS,GAAG,OAAO;AAAA,EACnD,SAAS,OAAO;AACd,UAAM,UAAU,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK;AACrE,UAAM,IAAI;AAAA,MACR,sCAAsC,OAAO,+DAExB,OAAO;AAAA,IAC9B;AAAA,EACF;AACF;AAeO,SAAS,cAAc,SAAiB,UAA+C;AAC5F,MAAI;AACF,OAAG,WAAW,OAAO;AACrB,WAAO,EAAE,SAAS,KAAK;AAAA,EACzB,SAAS,OAAO;AAEd,QAAI,iBAAiB,SAAS,UAAU,SAAS,MAAM,SAAS,UAAU;AACxE,aAAO,EAAE,SAAS,KAAK;AAAA,IACzB;AAEA,WAAO;AAAA,MACL,SAAS;AAAA,MACT,OAAO,iBAAiB,QAAQ,QAAQ,IAAI,MAAM,OAAO,KAAK,CAAC;AAAA,IACjE;AAAA,EACF;AACF;AASO,SAAS,iBAAiB,SAAiB,OAAuB;AACvE,SAAO,KAAK,KAAK,SAAS,GAAG,KAAK,MAAM;AAC1C;;;AF7DA,SAAS,iBAAiB,UAAwB;AAChD,MAAI,SAAS,MAAM,UAAU;AAE3B,eAAW,MAAM;AACf,cAAQ,KAAK,QAAQ,KAAK,SAAS;AAAA,IACrC,GAAG,GAAG;AAAA,EACR,OAAO;AACL,YAAQ,KAAK,QAAQ;AAAA,EACvB;AACF;AAEA,IAAMA,UAAS,aAAa,uBAAuB;AAUnD,eAAe,OAAsB;AACnC,QAAM,QAAQ,QAAQ,KAAK,CAAC;AAC5B,QAAM,UAAU,QAAQ,IAAI,iBAAiB;AAE7C,MAAI,UAAU,UAAa,UAAU,IAAI;AACvC,IAAAA,QAAO,MAAM,wDAAwD;AACrE,UAAM,eAAe;AACrB,YAAQ,KAAK,CAAC;AAAA,EAChB;AAGA,QAAM,aAAa,IAAI,WAAW,OAAO;AACzC,QAAM,WAAW,MAAM,eAAe,QAAW,OAAO;AAGxD,QAAM,UAAU;AAAA,IACd,WAAW,SAAS;AAAA;AAAA,IACpB;AAAA,EACF;AAEA,MAAI;AACF,iBAAa,SAAS,QAAQ,GAAG;AAAA,EACnC,SAAS,OAAO;AAEd,IAAAA,QAAO;AAAA,MACL,EAAE,OAAO,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK,EAAE;AAAA,MAChE;AAAA,IACF;AACA,UAAM,eAAe;AACrB,YAAQ,KAAK,CAAC;AAAA,EAChB;AAGA,UAAQ,GAAG,WAAW,MAAM;AAC1B,IAAAA,QAAO,KAAK,EAAE,MAAM,GAAG,kCAAkC;AACzD,eAAW,UAAU,OAAO;AAAA,MAC1B,QAAQ;AAAA,MACR,SAAS;AAAA,IACX,CAAC;AAGD,UAAM,eAAe,cAAc,SAAS,SAAS;AACrD,QAAI,CAAC,aAAa,WAAW,aAAa,UAAU,QAAW;AAC7D,MAAAA,QAAO;AAAA,QACL,EAAE,OAAO,OAAO,aAAa,MAAM,QAAQ;AAAA,QAC3C;AAAA,MACF;AAAA,IACF;AAGA,SAAK,eAAe,EAAE,QAAQ,MAAM,QAAQ,KAAK,CAAC,CAAC;AAAA,EACrD,CAAC;AAGD,QAAM,YAAY,MAAM,SAAS,OAAO,KAAK;AAC7C,QAAM,SAAS,IAAI;AAAA,IACjB;AAAA,IACA,SAAS;AAAA,IACT,SAAS;AAAA,IACT,SAAS;AAAA,IACT,SAAS;AAAA,IACT,UAAU;AAAA,EACZ;AAEA,MAAI;AACF,UAAM,OAAO,WAAW,KAAK;AAG7B,UAAM,iBAAiB,cAAc,SAAS,SAAS;AACvD,QAAI,CAAC,eAAe,WAAW,eAAe,UAAU,QAAW;AACjE,MAAAA,QAAO;AAAA,QACL,EAAE,OAAO,OAAO,eAAe,MAAM,QAAQ;AAAA,QAC7C;AAAA,MACF;AAAA,IACF;AAEA,IAAAA,QAAO,KAAK,EAAE,MAAM,GAAG,4BAA4B;AACnD,UAAM,gBAAgB,QAAQ;AAC9B,UAAM,eAAe;AACrB,qBAAiB,CAAC;AAAA,EACpB,SAAS,OAAO;AAEd,IAAAA,QAAO;AAAA,MACL,EAAE,OAAO,OAAO,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK,EAAE;AAAA,MACvE;AAAA,IACF;AAGA,UAAM,iBAAiB,cAAc,SAAS,SAAS;AACvD,QAAI,CAAC,eAAe,WAAW,eAAe,UAAU,QAAW;AACjE,MAAAA,QAAO;AAAA,QACL,EAAE,OAAO,OAAO,eAAe,MAAM,QAAQ;AAAA,QAC7C;AAAA,MACF;AAAA,IACF;AAEA,UAAM,gBAAgB,QAAQ;AAC9B,UAAM,eAAe;AACrB,qBAAiB,CAAC;AAAA,EACpB;AACF;AAEA,KAAK,EAAE,MAAM,OAAO,UAAmB;AACrC,EAAAA,QAAO;AAAA,IACL,EAAE,OAAO,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK,EAAE;AAAA,IAChE;AAAA,EACF;AACA,QAAM,eAAe;AACrB,mBAAiB,CAAC;AACpB,CAAC;","names":["logger"]}
1
+ {"version":3,"sources":["../../src/workers/background-worker-cli.ts","../../src/workers/background-worker.ts","../../src/workers/pid-file.ts"],"sourcesContent":["#!/usr/bin/env node\nimport { platform } from 'os';\nimport { BackgroundWorker } from './background-worker.js';\nimport { writePidFile, deletePidFile, buildPidFilePath } from './pid-file.js';\nimport { createLogger, shutdownLogger } from '../logging/index.js';\nimport { createServices, destroyServices } from '../services/index.js';\nimport { JobService } from '../services/job.service.js';\n\n/**\n * Force exit the process to avoid ONNX runtime mutex crash on macOS.\n *\n * On macOS, the ONNX runtime (used by transformers.js for embeddings) has a known\n * bug where static mutex cleanup fails during process exit, causing a crash with:\n * \"mutex lock failed: Invalid argument\"\n *\n * This doesn't affect job completion - all work is done and persisted before exit.\n * Using SIGKILL bypasses the problematic cleanup code.\n *\n * See: https://github.com/microsoft/onnxruntime/issues/24579\n */\nfunction forceExitOnMacOS(exitCode: number): void {\n if (platform() === 'darwin') {\n // Give time for any pending I/O to flush\n setTimeout(() => {\n process.kill(process.pid, 'SIGKILL');\n }, 100);\n } else {\n process.exit(exitCode);\n }\n}\n\nconst logger = createLogger('background-worker-cli');\n\n/**\n * Background worker CLI entry point\n *\n * Usage: background-worker-cli <job-id>\n *\n * This process runs detached from the parent and executes a single job.\n */\n\nasync function main(): Promise<void> {\n const jobId = process.argv[2];\n const dataDir = process.env['BLUERA_DATA_DIR'];\n\n if (jobId === undefined || jobId === '') {\n logger.error('Job ID required. Usage: background-worker-cli <job-id>');\n await shutdownLogger();\n process.exit(1);\n }\n\n // Initialize services\n const jobService = new JobService(dataDir);\n const services = await createServices(undefined, dataDir);\n\n // Write PID file for job cancellation - CRITICAL: must succeed or job cannot be cancelled\n const pidFile = buildPidFilePath(\n jobService['jobsDir'], // Access private field for PID path\n jobId\n );\n\n try {\n writePidFile(pidFile, process.pid);\n } catch (error) {\n // CRITICAL: Cannot proceed without PID file - job would be uncancellable\n logger.error(\n { error: error instanceof Error ? error.message : String(error) },\n 'Failed to write PID file'\n );\n await shutdownLogger();\n process.exit(1);\n }\n\n // Handle SIGTERM for graceful shutdown\n process.on('SIGTERM', () => {\n logger.info({ jobId }, 'Received SIGTERM, cancelling job');\n jobService.updateJob(jobId, {\n status: 'cancelled',\n message: 'Job cancelled by user',\n });\n\n // Clean up PID file (best-effort - don't block shutdown)\n const deleteResult = deletePidFile(pidFile, 'sigterm');\n if (!deleteResult.success && deleteResult.error !== undefined) {\n logger.warn(\n { jobId, error: deleteResult.error.message },\n 'Could not remove PID file during SIGTERM'\n );\n }\n\n // Flush logs before exit (best-effort, don't await in signal handler)\n void shutdownLogger().finally(() => process.exit(0));\n });\n\n // Load config and create worker\n const appConfig = await services.config.load();\n const worker = new BackgroundWorker(\n jobService,\n services.store,\n services.index,\n services.lance,\n services.embeddings,\n appConfig.crawl\n );\n\n try {\n await worker.executeJob(jobId);\n\n // Clean up PID file on success (best-effort - don't change exit code)\n const successCleanup = deletePidFile(pidFile, 'success');\n if (!successCleanup.success && successCleanup.error !== undefined) {\n logger.warn(\n { jobId, error: successCleanup.error.message },\n 'Could not remove PID file after success'\n );\n }\n\n logger.info({ jobId }, 'Job completed successfully');\n await destroyServices(services);\n await shutdownLogger();\n forceExitOnMacOS(0);\n } catch (error) {\n // Job service already updated with failure status in BackgroundWorker\n logger.error(\n { jobId, error: error instanceof Error ? error.message : String(error) },\n 'Job failed'\n );\n\n // Clean up PID file on failure (best-effort - exit code reflects job failure)\n const failureCleanup = deletePidFile(pidFile, 'failure');\n if (!failureCleanup.success && failureCleanup.error !== undefined) {\n logger.warn(\n { jobId, error: failureCleanup.error.message },\n 'Could not remove PID file after failure'\n );\n }\n\n await destroyServices(services);\n await shutdownLogger();\n forceExitOnMacOS(1);\n }\n}\n\nmain().catch(async (error: unknown) => {\n logger.error(\n { error: error instanceof Error ? error.message : String(error) },\n 'Fatal error in background worker'\n );\n await shutdownLogger();\n forceExitOnMacOS(1);\n});\n","import { createHash } from 'node:crypto';\nimport {\n IntelligentCrawler,\n type CrawlConfig,\n type CrawlProgress,\n} from '../crawl/intelligent-crawler.js';\nimport { createLogger } from '../logging/index.js';\nimport { IndexService } from '../services/index.service.js';\nimport { JobService } from '../services/job.service.js';\nimport { StoreService } from '../services/store.service.js';\nimport { createStoreId, createDocumentId } from '../types/brands.js';\nimport type { EmbeddingEngine } from '../db/embeddings.js';\nimport type { LanceStore } from '../db/lance.js';\nimport type { Document } from '../types/document.js';\nimport type { Job } from '../types/job.js';\n\nconst logger = createLogger('background-worker');\n\n/**\n * Calculate index progress as a percentage, handling division by zero.\n * @param current - Current number of items processed\n * @param total - Total number of items (may be 0)\n * @param scale - Scale factor for progress (default 100 for 0-100%)\n * @returns Progress value, or 0 if total is 0\n */\nexport function calculateIndexProgress(\n current: number,\n total: number,\n scale: number = 100\n): number {\n if (total === 0) return 0;\n return (current / total) * scale;\n}\n\nexport class BackgroundWorker {\n constructor(\n private readonly jobService: JobService,\n private readonly storeService: StoreService,\n private readonly indexService: IndexService,\n private readonly lanceStore: LanceStore,\n private readonly embeddingEngine: EmbeddingEngine,\n private readonly crawlConfig?: CrawlConfig\n ) {}\n\n /**\n * Execute a job based on its type\n */\n async executeJob(jobId: string): Promise<void> {\n const job = this.jobService.getJob(jobId);\n\n if (!job) {\n throw new Error(`Job ${jobId} not found`);\n }\n\n try {\n logger.info({ jobId, type: job.type }, 'Starting job execution');\n\n // Update to running status\n this.jobService.updateJob(jobId, {\n status: 'running',\n message: `Starting ${job.type} operation...`,\n progress: 0,\n details: { startedAt: new Date().toISOString() },\n });\n\n // Execute based on job type\n switch (job.type) {\n case 'clone':\n await this.executeCloneJob(job);\n break;\n case 'index':\n await this.executeIndexJob(job);\n break;\n case 'crawl':\n await this.executeCrawlJob(job);\n break;\n default:\n throw new Error(`Unknown job type: ${String(job.type)}`);\n }\n\n // Mark as completed\n this.jobService.updateJob(jobId, {\n status: 'completed',\n progress: 100,\n message: `${job.type} operation completed successfully`,\n details: { completedAt: new Date().toISOString() },\n });\n } catch (error) {\n logger.error(\n { jobId, error: error instanceof Error ? error.message : String(error) },\n 'Job failed'\n );\n\n // Mark as failed\n const errorDetails: Record<string, unknown> = {\n completedAt: new Date().toISOString(),\n };\n if (error instanceof Error && error.stack !== undefined) {\n errorDetails['error'] = error.stack;\n } else {\n errorDetails['error'] = String(error);\n }\n this.jobService.updateJob(jobId, {\n status: 'failed',\n message: error instanceof Error ? error.message : 'Unknown error',\n details: errorDetails,\n });\n throw error;\n }\n }\n\n /**\n * Execute a clone job (git clone + initial indexing)\n */\n private async executeCloneJob(job: Job): Promise<void> {\n const { storeId } = job.details;\n\n if (storeId === undefined || typeof storeId !== 'string') {\n throw new Error('Store ID required for clone job');\n }\n\n // Get the store\n const store = await this.storeService.get(createStoreId(storeId));\n if (!store) {\n throw new Error(`Store ${storeId} not found`);\n }\n\n // Clone is already done by the time the job is created\n // (happens in StoreService.create), so we just need to index\n\n // Update progress - cloning considered done (30%)\n this.jobService.updateJob(job.id, {\n status: 'running',\n message: 'Repository cloned, starting indexing...',\n progress: 30,\n });\n\n // Initialize LanceStore with dimensions before indexing\n this.lanceStore.setDimensions(await this.embeddingEngine.ensureDimensions());\n await this.lanceStore.initialize(store.id);\n\n // Index the repository with progress updates\n const result = await this.indexService.indexStore(\n store,\n (event: { type: string; current: number; total: number; message: string }) => {\n // Check if job was cancelled\n const currentJob = this.jobService.getJob(job.id);\n if (currentJob?.status === 'cancelled') {\n throw new Error('Job cancelled by user');\n }\n\n // Indexing is 70% of total progress (30-100%)\n const indexProgress = calculateIndexProgress(event.current, event.total, 70);\n const totalProgress = 30 + indexProgress;\n\n this.jobService.updateJob(job.id, {\n message: `Indexed ${String(event.current)}/${String(event.total)} files`,\n progress: Math.min(99, totalProgress), // Cap at 99 until fully complete\n details: {\n filesProcessed: event.current,\n totalFiles: event.total,\n },\n });\n }\n );\n\n if (!result.success) {\n throw result.error;\n }\n }\n\n /**\n * Execute an index job (re-indexing existing store)\n */\n private async executeIndexJob(job: Job): Promise<void> {\n const { storeId } = job.details;\n\n if (storeId === undefined || typeof storeId !== 'string') {\n throw new Error('Store ID required for index job');\n }\n\n // Get the store\n const store = await this.storeService.getByIdOrName(createStoreId(storeId));\n if (!store) {\n throw new Error(`Store ${storeId} not found`);\n }\n\n // Initialize LanceStore with dimensions before indexing\n this.lanceStore.setDimensions(await this.embeddingEngine.ensureDimensions());\n await this.lanceStore.initialize(store.id);\n\n // Index with progress updates\n const result = await this.indexService.indexStore(\n store,\n (event: { type: string; current: number; total: number; message: string }) => {\n // Check if job was cancelled\n const currentJob = this.jobService.getJob(job.id);\n if (currentJob?.status === 'cancelled') {\n throw new Error('Job cancelled by user');\n }\n\n const progress = calculateIndexProgress(event.current, event.total);\n\n this.jobService.updateJob(job.id, {\n message: `Indexed ${String(event.current)}/${String(event.total)} files`,\n progress: Math.min(99, progress), // Cap at 99 until fully complete\n details: {\n filesProcessed: event.current,\n totalFiles: event.total,\n },\n });\n }\n );\n\n if (!result.success) {\n throw result.error;\n }\n }\n\n /**\n * Execute a crawl job (web crawling + indexing)\n */\n private async executeCrawlJob(job: Job): Promise<void> {\n const { storeId, url, crawlInstruction, extractInstruction, maxPages, simple, useHeadless } =\n job.details;\n\n if (storeId === undefined || typeof storeId !== 'string') {\n throw new Error('Store ID required for crawl job');\n }\n if (url === undefined || typeof url !== 'string') {\n throw new Error('URL required for crawl job');\n }\n\n // Get the store\n const store = await this.storeService.get(createStoreId(storeId));\n if (store?.type !== 'web') {\n throw new Error(`Web store ${storeId} not found`);\n }\n\n const resolvedMaxPages = typeof maxPages === 'number' ? maxPages : 50;\n const crawler = new IntelligentCrawler(this.crawlConfig);\n\n // Listen for progress events\n crawler.on('progress', (progress: CrawlProgress) => {\n // Check if job was cancelled - just return early, for-await loop will throw and finally will cleanup\n const currentJob = this.jobService.getJob(job.id);\n if (currentJob?.status === 'cancelled') {\n return;\n }\n\n // Crawling is 80% of total progress (0-80%)\n const crawlProgress = (progress.pagesVisited / resolvedMaxPages) * 80;\n\n this.jobService.updateJob(job.id, {\n message:\n progress.message ??\n `Crawling page ${String(progress.pagesVisited)}/${String(resolvedMaxPages)}`,\n progress: Math.min(80, crawlProgress),\n details: { pagesCrawled: progress.pagesVisited },\n });\n });\n\n try {\n this.lanceStore.setDimensions(await this.embeddingEngine.ensureDimensions());\n await this.lanceStore.initialize(store.id);\n const docs: Document[] = [];\n\n // Build crawl options, only including defined values\n const crawlOptions: {\n maxPages: number;\n simple: boolean;\n useHeadless: boolean;\n crawlInstruction?: string;\n extractInstruction?: string;\n } = {\n maxPages: resolvedMaxPages,\n simple: simple ?? false,\n useHeadless: useHeadless ?? true, // Default to headless for reliability\n };\n if (crawlInstruction !== undefined) {\n crawlOptions.crawlInstruction = crawlInstruction;\n }\n if (extractInstruction !== undefined) {\n crawlOptions.extractInstruction = extractInstruction;\n }\n\n // Crawl pages using IntelligentCrawler\n for await (const result of crawler.crawl(url, crawlOptions)) {\n // Check cancellation between pages\n const currentJob = this.jobService.getJob(job.id);\n if (currentJob?.status === 'cancelled') {\n throw new Error('Job cancelled by user');\n }\n\n // Embed and index the content (use extracted if available, otherwise markdown)\n const contentToEmbed = result.extracted ?? result.markdown;\n const vector = await this.embeddingEngine.embedDocument(contentToEmbed);\n\n docs.push({\n id: createDocumentId(`${store.id}-${createHash('md5').update(result.url).digest('hex')}`),\n content: contentToEmbed,\n vector,\n metadata: {\n type: 'web',\n storeId: store.id,\n url: result.url,\n title: result.title,\n extracted: result.extracted !== undefined,\n depth: result.depth,\n indexedAt: new Date().toISOString(),\n },\n });\n }\n\n // Index all documents (remaining 20%)\n if (docs.length > 0) {\n this.jobService.updateJob(job.id, {\n message: 'Indexing crawled documents...',\n progress: 85,\n });\n\n // Clear existing documents to prevent duplicates on re-crawl\n await this.lanceStore.clearAllDocuments(store.id);\n await this.lanceStore.addDocuments(store.id, docs);\n // Create FTS index for full-text search\n await this.lanceStore.createFtsIndex(store.id);\n }\n\n this.jobService.updateJob(job.id, {\n message: `Crawled and indexed ${String(docs.length)} pages`,\n progress: 100,\n details: { pagesCrawled: docs.length },\n });\n } finally {\n await crawler.stop();\n }\n }\n}\n","import fs from 'fs';\nimport path from 'path';\n\n/**\n * Result of a PID file delete operation.\n * Delete operations are best-effort and should not throw.\n */\nexport interface PidFileResult {\n success: boolean;\n error?: Error;\n}\n\n/**\n * Context for PID file deletion - indicates when the delete is happening.\n * Used for logging/debugging purposes.\n */\nexport type PidFileDeleteContext = 'sigterm' | 'success' | 'failure';\n\n/**\n * Write PID file - CRITICAL operation that must succeed.\n *\n * If the PID file cannot be written, the job cannot be cancelled through\n * the job management system. This is a critical failure and the job\n * should not proceed.\n *\n * @param pidFile - Absolute path to the PID file\n * @param pid - Process ID to write\n * @throws Error if PID file cannot be written\n */\nexport function writePidFile(pidFile: string, pid: number): void {\n try {\n fs.writeFileSync(pidFile, pid.toString(), 'utf-8');\n } catch (error) {\n const message = error instanceof Error ? error.message : String(error);\n throw new Error(\n `CRITICAL: Failed to write PID file ${pidFile}. ` +\n `Job cannot be cancelled without PID file. ` +\n `Original error: ${message}`\n );\n }\n}\n\n/**\n * Delete PID file - best-effort cleanup during shutdown.\n *\n * This operation should NEVER throw. During process shutdown (SIGTERM,\n * job success, job failure), failing to delete a PID file should not\n * prevent the process from exiting cleanly.\n *\n * Stale PID files are cleaned up by JobService.cleanupOldJobs().\n *\n * @param pidFile - Absolute path to the PID file\n * @param _context - Context indicating when the delete is happening (for future logging)\n * @returns Result indicating success or failure with error details\n */\nexport function deletePidFile(pidFile: string, _context: PidFileDeleteContext): PidFileResult {\n try {\n fs.unlinkSync(pidFile);\n return { success: true };\n } catch (error) {\n // ENOENT = file doesn't exist - that's success (nothing to delete)\n if (error instanceof Error && 'code' in error && error.code === 'ENOENT') {\n return { success: true };\n }\n // Any other error = failure (permission denied, etc.)\n return {\n success: false,\n error: error instanceof Error ? error : new Error(String(error)),\n };\n }\n}\n\n/**\n * Build the path to a PID file for a given job.\n *\n * @param jobsDir - Directory where job files are stored\n * @param jobId - Job identifier\n * @returns Absolute path to the PID file\n */\nexport function buildPidFilePath(jobsDir: string, jobId: string): string {\n return path.join(jobsDir, `${jobId}.pid`);\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;;AACA,SAAS,gBAAgB;;;ACDzB,SAAS,kBAAkB;AAgB3B,IAAM,SAAS,aAAa,mBAAmB;AASxC,SAAS,uBACd,SACA,OACA,QAAgB,KACR;AACR,MAAI,UAAU,EAAG,QAAO;AACxB,SAAQ,UAAU,QAAS;AAC7B;AAEO,IAAM,mBAAN,MAAuB;AAAA,EAC5B,YACmB,YACA,cACA,cACA,YACA,iBACA,aACjB;AANiB;AACA;AACA;AACA;AACA;AACA;AAAA,EAChB;AAAA;AAAA;AAAA;AAAA,EAKH,MAAM,WAAW,OAA8B;AAC7C,UAAM,MAAM,KAAK,WAAW,OAAO,KAAK;AAExC,QAAI,CAAC,KAAK;AACR,YAAM,IAAI,MAAM,OAAO,KAAK,YAAY;AAAA,IAC1C;AAEA,QAAI;AACF,aAAO,KAAK,EAAE,OAAO,MAAM,IAAI,KAAK,GAAG,wBAAwB;AAG/D,WAAK,WAAW,UAAU,OAAO;AAAA,QAC/B,QAAQ;AAAA,QACR,SAAS,YAAY,IAAI,IAAI;AAAA,QAC7B,UAAU;AAAA,QACV,SAAS,EAAE,YAAW,oBAAI,KAAK,GAAE,YAAY,EAAE;AAAA,MACjD,CAAC;AAGD,cAAQ,IAAI,MAAM;AAAA,QAChB,KAAK;AACH,gBAAM,KAAK,gBAAgB,GAAG;AAC9B;AAAA,QACF,KAAK;AACH,gBAAM,KAAK,gBAAgB,GAAG;AAC9B;AAAA,QACF,KAAK;AACH,gBAAM,KAAK,gBAAgB,GAAG;AAC9B;AAAA,QACF;AACE,gBAAM,IAAI,MAAM,qBAAqB,OAAO,IAAI,IAAI,CAAC,EAAE;AAAA,MAC3D;AAGA,WAAK,WAAW,UAAU,OAAO;AAAA,QAC/B,QAAQ;AAAA,QACR,UAAU;AAAA,QACV,SAAS,GAAG,IAAI,IAAI;AAAA,QACpB,SAAS,EAAE,cAAa,oBAAI,KAAK,GAAE,YAAY,EAAE;AAAA,MACnD,CAAC;AAAA,IACH,SAAS,OAAO;AACd,aAAO;AAAA,QACL,EAAE,OAAO,OAAO,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK,EAAE;AAAA,QACvE;AAAA,MACF;AAGA,YAAM,eAAwC;AAAA,QAC5C,cAAa,oBAAI,KAAK,GAAE,YAAY;AAAA,MACtC;AACA,UAAI,iBAAiB,SAAS,MAAM,UAAU,QAAW;AACvD,qBAAa,OAAO,IAAI,MAAM;AAAA,MAChC,OAAO;AACL,qBAAa,OAAO,IAAI,OAAO,KAAK;AAAA,MACtC;AACA,WAAK,WAAW,UAAU,OAAO;AAAA,QAC/B,QAAQ;AAAA,QACR,SAAS,iBAAiB,QAAQ,MAAM,UAAU;AAAA,QAClD,SAAS;AAAA,MACX,CAAC;AACD,YAAM;AAAA,IACR;AAAA,EACF;AAAA;AAAA;AAAA;AAAA,EAKA,MAAc,gBAAgB,KAAyB;AACrD,UAAM,EAAE,QAAQ,IAAI,IAAI;AAExB,QAAI,YAAY,UAAa,OAAO,YAAY,UAAU;AACxD,YAAM,IAAI,MAAM,iCAAiC;AAAA,IACnD;AAGA,UAAM,QAAQ,MAAM,KAAK,aAAa,IAAI,cAAc,OAAO,CAAC;AAChE,QAAI,CAAC,OAAO;AACV,YAAM,IAAI,MAAM,SAAS,OAAO,YAAY;AAAA,IAC9C;AAMA,SAAK,WAAW,UAAU,IAAI,IAAI;AAAA,MAChC,QAAQ;AAAA,MACR,SAAS;AAAA,MACT,UAAU;AAAA,IACZ,CAAC;AAGD,SAAK,WAAW,cAAc,MAAM,KAAK,gBAAgB,iBAAiB,CAAC;AAC3E,UAAM,KAAK,WAAW,WAAW,MAAM,EAAE;AAGzC,UAAM,SAAS,MAAM,KAAK,aAAa;AAAA,MACrC;AAAA,MACA,CAAC,UAA6E;AAE5E,cAAM,aAAa,KAAK,WAAW,OAAO,IAAI,EAAE;AAChD,YAAI,YAAY,WAAW,aAAa;AACtC,gBAAM,IAAI,MAAM,uBAAuB;AAAA,QACzC;AAGA,cAAM,gBAAgB,uBAAuB,MAAM,SAAS,MAAM,OAAO,EAAE;AAC3E,cAAM,gBAAgB,KAAK;AAE3B,aAAK,WAAW,UAAU,IAAI,IAAI;AAAA,UAChC,SAAS,WAAW,OAAO,MAAM,OAAO,CAAC,IAAI,OAAO,MAAM,KAAK,CAAC;AAAA,UAChE,UAAU,KAAK,IAAI,IAAI,aAAa;AAAA;AAAA,UACpC,SAAS;AAAA,YACP,gBAAgB,MAAM;AAAA,YACtB,YAAY,MAAM;AAAA,UACpB;AAAA,QACF,CAAC;AAAA,MACH;AAAA,IACF;AAEA,QAAI,CAAC,OAAO,SAAS;AACnB,YAAM,OAAO;AAAA,IACf;AAAA,EACF;AAAA;AAAA;AAAA;AAAA,EAKA,MAAc,gBAAgB,KAAyB;AACrD,UAAM,EAAE,QAAQ,IAAI,IAAI;AAExB,QAAI,YAAY,UAAa,OAAO,YAAY,UAAU;AACxD,YAAM,IAAI,MAAM,iCAAiC;AAAA,IACnD;AAGA,UAAM,QAAQ,MAAM,KAAK,aAAa,cAAc,cAAc,OAAO,CAAC;AAC1E,QAAI,CAAC,OAAO;AACV,YAAM,IAAI,MAAM,SAAS,OAAO,YAAY;AAAA,IAC9C;AAGA,SAAK,WAAW,cAAc,MAAM,KAAK,gBAAgB,iBAAiB,CAAC;AAC3E,UAAM,KAAK,WAAW,WAAW,MAAM,EAAE;AAGzC,UAAM,SAAS,MAAM,KAAK,aAAa;AAAA,MACrC;AAAA,MACA,CAAC,UAA6E;AAE5E,cAAM,aAAa,KAAK,WAAW,OAAO,IAAI,EAAE;AAChD,YAAI,YAAY,WAAW,aAAa;AACtC,gBAAM,IAAI,MAAM,uBAAuB;AAAA,QACzC;AAEA,cAAM,WAAW,uBAAuB,MAAM,SAAS,MAAM,KAAK;AAElE,aAAK,WAAW,UAAU,IAAI,IAAI;AAAA,UAChC,SAAS,WAAW,OAAO,MAAM,OAAO,CAAC,IAAI,OAAO,MAAM,KAAK,CAAC;AAAA,UAChE,UAAU,KAAK,IAAI,IAAI,QAAQ;AAAA;AAAA,UAC/B,SAAS;AAAA,YACP,gBAAgB,MAAM;AAAA,YACtB,YAAY,MAAM;AAAA,UACpB;AAAA,QACF,CAAC;AAAA,MACH;AAAA,IACF;AAEA,QAAI,CAAC,OAAO,SAAS;AACnB,YAAM,OAAO;AAAA,IACf;AAAA,EACF;AAAA;AAAA;AAAA;AAAA,EAKA,MAAc,gBAAgB,KAAyB;AACrD,UAAM,EAAE,SAAS,KAAK,kBAAkB,oBAAoB,UAAU,QAAQ,YAAY,IACxF,IAAI;AAEN,QAAI,YAAY,UAAa,OAAO,YAAY,UAAU;AACxD,YAAM,IAAI,MAAM,iCAAiC;AAAA,IACnD;AACA,QAAI,QAAQ,UAAa,OAAO,QAAQ,UAAU;AAChD,YAAM,IAAI,MAAM,4BAA4B;AAAA,IAC9C;AAGA,UAAM,QAAQ,MAAM,KAAK,aAAa,IAAI,cAAc,OAAO,CAAC;AAChE,QAAI,OAAO,SAAS,OAAO;AACzB,YAAM,IAAI,MAAM,aAAa,OAAO,YAAY;AAAA,IAClD;AAEA,UAAM,mBAAmB,OAAO,aAAa,WAAW,WAAW;AACnE,UAAM,UAAU,IAAI,mBAAmB,KAAK,WAAW;AAGvD,YAAQ,GAAG,YAAY,CAAC,aAA4B;AAElD,YAAM,aAAa,KAAK,WAAW,OAAO,IAAI,EAAE;AAChD,UAAI,YAAY,WAAW,aAAa;AACtC;AAAA,MACF;AAGA,YAAM,gBAAiB,SAAS,eAAe,mBAAoB;AAEnE,WAAK,WAAW,UAAU,IAAI,IAAI;AAAA,QAChC,SACE,SAAS,WACT,iBAAiB,OAAO,SAAS,YAAY,CAAC,IAAI,OAAO,gBAAgB,CAAC;AAAA,QAC5E,UAAU,KAAK,IAAI,IAAI,aAAa;AAAA,QACpC,SAAS,EAAE,cAAc,SAAS,aAAa;AAAA,MACjD,CAAC;AAAA,IACH,CAAC;AAED,QAAI;AACF,WAAK,WAAW,cAAc,MAAM,KAAK,gBAAgB,iBAAiB,CAAC;AAC3E,YAAM,KAAK,WAAW,WAAW,MAAM,EAAE;AACzC,YAAM,OAAmB,CAAC;AAG1B,YAAM,eAMF;AAAA,QACF,UAAU;AAAA,QACV,QAAQ,UAAU;AAAA,QAClB,aAAa,eAAe;AAAA;AAAA,MAC9B;AACA,UAAI,qBAAqB,QAAW;AAClC,qBAAa,mBAAmB;AAAA,MAClC;AACA,UAAI,uBAAuB,QAAW;AACpC,qBAAa,qBAAqB;AAAA,MACpC;AAGA,uBAAiB,UAAU,QAAQ,MAAM,KAAK,YAAY,GAAG;AAE3D,cAAM,aAAa,KAAK,WAAW,OAAO,IAAI,EAAE;AAChD,YAAI,YAAY,WAAW,aAAa;AACtC,gBAAM,IAAI,MAAM,uBAAuB;AAAA,QACzC;AAGA,cAAM,iBAAiB,OAAO,aAAa,OAAO;AAClD,cAAM,SAAS,MAAM,KAAK,gBAAgB,cAAc,cAAc;AAEtE,aAAK,KAAK;AAAA,UACR,IAAI,iBAAiB,GAAG,MAAM,EAAE,IAAI,WAAW,KAAK,EAAE,OAAO,OAAO,GAAG,EAAE,OAAO,KAAK,CAAC,EAAE;AAAA,UACxF,SAAS;AAAA,UACT;AAAA,UACA,UAAU;AAAA,YACR,MAAM;AAAA,YACN,SAAS,MAAM;AAAA,YACf,KAAK,OAAO;AAAA,YACZ,OAAO,OAAO;AAAA,YACd,WAAW,OAAO,cAAc;AAAA,YAChC,OAAO,OAAO;AAAA,YACd,YAAW,oBAAI,KAAK,GAAE,YAAY;AAAA,UACpC;AAAA,QACF,CAAC;AAAA,MACH;AAGA,UAAI,KAAK,SAAS,GAAG;AACnB,aAAK,WAAW,UAAU,IAAI,IAAI;AAAA,UAChC,SAAS;AAAA,UACT,UAAU;AAAA,QACZ,CAAC;AAGD,cAAM,KAAK,WAAW,kBAAkB,MAAM,EAAE;AAChD,cAAM,KAAK,WAAW,aAAa,MAAM,IAAI,IAAI;AAEjD,cAAM,KAAK,WAAW,eAAe,MAAM,EAAE;AAAA,MAC/C;AAEA,WAAK,WAAW,UAAU,IAAI,IAAI;AAAA,QAChC,SAAS,uBAAuB,OAAO,KAAK,MAAM,CAAC;AAAA,QACnD,UAAU;AAAA,QACV,SAAS,EAAE,cAAc,KAAK,OAAO;AAAA,MACvC,CAAC;AAAA,IACH,UAAE;AACA,YAAM,QAAQ,KAAK;AAAA,IACrB;AAAA,EACF;AACF;;;ACjVA,OAAO,QAAQ;AACf,OAAO,UAAU;AA4BV,SAAS,aAAa,SAAiB,KAAmB;AAC/D,MAAI;AACF,OAAG,cAAc,SAAS,IAAI,SAAS,GAAG,OAAO;AAAA,EACnD,SAAS,OAAO;AACd,UAAM,UAAU,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK;AACrE,UAAM,IAAI;AAAA,MACR,sCAAsC,OAAO,+DAExB,OAAO;AAAA,IAC9B;AAAA,EACF;AACF;AAeO,SAAS,cAAc,SAAiB,UAA+C;AAC5F,MAAI;AACF,OAAG,WAAW,OAAO;AACrB,WAAO,EAAE,SAAS,KAAK;AAAA,EACzB,SAAS,OAAO;AAEd,QAAI,iBAAiB,SAAS,UAAU,SAAS,MAAM,SAAS,UAAU;AACxE,aAAO,EAAE,SAAS,KAAK;AAAA,IACzB;AAEA,WAAO;AAAA,MACL,SAAS;AAAA,MACT,OAAO,iBAAiB,QAAQ,QAAQ,IAAI,MAAM,OAAO,KAAK,CAAC;AAAA,IACjE;AAAA,EACF;AACF;AASO,SAAS,iBAAiB,SAAiB,OAAuB;AACvE,SAAO,KAAK,KAAK,SAAS,GAAG,KAAK,MAAM;AAC1C;;;AF7DA,SAAS,iBAAiB,UAAwB;AAChD,MAAI,SAAS,MAAM,UAAU;AAE3B,eAAW,MAAM;AACf,cAAQ,KAAK,QAAQ,KAAK,SAAS;AAAA,IACrC,GAAG,GAAG;AAAA,EACR,OAAO;AACL,YAAQ,KAAK,QAAQ;AAAA,EACvB;AACF;AAEA,IAAMA,UAAS,aAAa,uBAAuB;AAUnD,eAAe,OAAsB;AACnC,QAAM,QAAQ,QAAQ,KAAK,CAAC;AAC5B,QAAM,UAAU,QAAQ,IAAI,iBAAiB;AAE7C,MAAI,UAAU,UAAa,UAAU,IAAI;AACvC,IAAAA,QAAO,MAAM,wDAAwD;AACrE,UAAM,eAAe;AACrB,YAAQ,KAAK,CAAC;AAAA,EAChB;AAGA,QAAM,aAAa,IAAI,WAAW,OAAO;AACzC,QAAM,WAAW,MAAM,eAAe,QAAW,OAAO;AAGxD,QAAM,UAAU;AAAA,IACd,WAAW,SAAS;AAAA;AAAA,IACpB;AAAA,EACF;AAEA,MAAI;AACF,iBAAa,SAAS,QAAQ,GAAG;AAAA,EACnC,SAAS,OAAO;AAEd,IAAAA,QAAO;AAAA,MACL,EAAE,OAAO,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK,EAAE;AAAA,MAChE;AAAA,IACF;AACA,UAAM,eAAe;AACrB,YAAQ,KAAK,CAAC;AAAA,EAChB;AAGA,UAAQ,GAAG,WAAW,MAAM;AAC1B,IAAAA,QAAO,KAAK,EAAE,MAAM,GAAG,kCAAkC;AACzD,eAAW,UAAU,OAAO;AAAA,MAC1B,QAAQ;AAAA,MACR,SAAS;AAAA,IACX,CAAC;AAGD,UAAM,eAAe,cAAc,SAAS,SAAS;AACrD,QAAI,CAAC,aAAa,WAAW,aAAa,UAAU,QAAW;AAC7D,MAAAA,QAAO;AAAA,QACL,EAAE,OAAO,OAAO,aAAa,MAAM,QAAQ;AAAA,QAC3C;AAAA,MACF;AAAA,IACF;AAGA,SAAK,eAAe,EAAE,QAAQ,MAAM,QAAQ,KAAK,CAAC,CAAC;AAAA,EACrD,CAAC;AAGD,QAAM,YAAY,MAAM,SAAS,OAAO,KAAK;AAC7C,QAAM,SAAS,IAAI;AAAA,IACjB;AAAA,IACA,SAAS;AAAA,IACT,SAAS;AAAA,IACT,SAAS;AAAA,IACT,SAAS;AAAA,IACT,UAAU;AAAA,EACZ;AAEA,MAAI;AACF,UAAM,OAAO,WAAW,KAAK;AAG7B,UAAM,iBAAiB,cAAc,SAAS,SAAS;AACvD,QAAI,CAAC,eAAe,WAAW,eAAe,UAAU,QAAW;AACjE,MAAAA,QAAO;AAAA,QACL,EAAE,OAAO,OAAO,eAAe,MAAM,QAAQ;AAAA,QAC7C;AAAA,MACF;AAAA,IACF;AAEA,IAAAA,QAAO,KAAK,EAAE,MAAM,GAAG,4BAA4B;AACnD,UAAM,gBAAgB,QAAQ;AAC9B,UAAM,eAAe;AACrB,qBAAiB,CAAC;AAAA,EACpB,SAAS,OAAO;AAEd,IAAAA,QAAO;AAAA,MACL,EAAE,OAAO,OAAO,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK,EAAE;AAAA,MACvE;AAAA,IACF;AAGA,UAAM,iBAAiB,cAAc,SAAS,SAAS;AACvD,QAAI,CAAC,eAAe,WAAW,eAAe,UAAU,QAAW;AACjE,MAAAA,QAAO;AAAA,QACL,EAAE,OAAO,OAAO,eAAe,MAAM,QAAQ;AAAA,QAC7C;AAAA,MACF;AAAA,IACF;AAEA,UAAM,gBAAgB,QAAQ;AAC9B,UAAM,eAAe;AACrB,qBAAiB,CAAC;AAAA,EACpB;AACF;AAEA,KAAK,EAAE,MAAM,OAAO,UAAmB;AACrC,EAAAA,QAAO;AAAA,IACL,EAAE,OAAO,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK,EAAE;AAAA,IAChE;AAAA,EACF;AACA,QAAM,eAAe;AACrB,mBAAiB,CAAC;AACpB,CAAC;","names":["logger"]}
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "bluera-knowledge",
3
- "version": "0.19.7",
3
+ "version": "0.21.0",
4
4
  "description": "CLI tool for managing knowledge stores with semantic search",
5
5
  "type": "module",
6
6
  "bin": {
@@ -18,7 +18,7 @@
18
18
  "test:run": "vitest run",
19
19
  "test:changed": "bash -c 'OUTPUT=$(vitest run --changed --reporter=dot --silent 2>&1); echo \"$OUTPUT\" | tail -5; if echo \"$OUTPUT\" | grep -qE \"Tests.*[0-9]+ failed\"; then exit 1; fi'",
20
20
  "test:changed:verbose": "vitest run --changed",
21
- "test:coverage": "bash -c 'OUTPUT=$(vitest run --coverage --reporter=dot 2>&1); echo \"$OUTPUT\" | grep -E \"(Test Files|Tests|Coverage|threshold|ERROR)\" | tail -10; if echo \"$OUTPUT\" | grep -qE \"Tests.*[0-9]+ failed\"; then exit 1; fi; if echo \"$OUTPUT\" | grep -qE \"ERROR.*threshold\"; then exit 1; fi'",
21
+ "test:coverage": "bash -c 'OUTPUT=$(vitest run --coverage --reporter=dot 2>&1); echo \"$OUTPUT\" | grep -E \"(Test Files|Tests|Coverage|threshold|ERROR|FAIL|AssertionError)\" | tail -20; if echo \"$OUTPUT\" | grep -qE \"Tests.*[0-9]+ failed\"; then echo \"$OUTPUT\" | grep -B5 -A5 \"AssertionError\\|FAIL.*integration\" | tail -30; exit 1; fi; if echo \"$OUTPUT\" | grep -qE \"ERROR.*threshold\"; then exit 1; fi'",
22
22
  "test:coverage:verbose": "vitest run --coverage",
23
23
  "format": "prettier --write \"src/**/*.ts\" \"tests/**/*.ts\"",
24
24
  "format:check": "prettier --check \"src/**/*.ts\" \"tests/**/*.ts\" --log-level warn && echo '✓ Format check passed'",