bluera-knowledge 0.12.7 → 0.12.9

package/dist/index.js

@@ -7,10 +7,10 @@ import {
   isWebStoreDefinition,
   runMCPServer,
   spawnBackgroundWorker
-} from "./chunk-PFBSZTP3.js";
+} from "./chunk-4ZBK7V54.js";
 import {
   IntelligentCrawler
-} from "./chunk-QCSFBMYW.js";
+} from "./chunk-6777ULXC.js";
 import {
   ASTParser,
   AdapterRegistry,
@@ -24,7 +24,7 @@ import {
   err,
   extractRepoName,
   ok
-} from "./chunk-C4SYGLAI.js";
+} from "./chunk-VTATT3IR.js";
 import "./chunk-HRQD3MPH.js";
 
 // src/index.ts

package/dist/mcp/server.js

@@ -1,8 +1,8 @@
 import {
   createMCPServer,
   runMCPServer
-} from "../chunk-PFBSZTP3.js";
-import "../chunk-C4SYGLAI.js";
+} from "../chunk-4ZBK7V54.js";
+import "../chunk-VTATT3IR.js";
 import "../chunk-HRQD3MPH.js";
 export {
   createMCPServer,

package/dist/workers/background-worker-cli.js

@@ -1,17 +1,21 @@
 #!/usr/bin/env node
 import {
   IntelligentCrawler
-} from "../chunk-QCSFBMYW.js";
+} from "../chunk-6777ULXC.js";
 import {
   JobService,
   createDocumentId,
   createLogger,
   createServices,
   createStoreId,
+  destroyServices,
   shutdownLogger
-} from "../chunk-C4SYGLAI.js";
+} from "../chunk-VTATT3IR.js";
 import "../chunk-HRQD3MPH.js";
 
+// src/workers/background-worker-cli.ts
+import { platform } from "os";
+
 // src/workers/background-worker.ts
 import { createHash } from "crypto";
 var logger = createLogger("background-worker");
@@ -276,6 +280,15 @@ function buildPidFilePath(jobsDir, jobId) {
 }
 
 // src/workers/background-worker-cli.ts
+function forceExitOnMacOS(exitCode) {
+  if (platform() === "darwin") {
+    setTimeout(() => {
+      process.kill(process.pid, "SIGKILL");
+    }, 100);
+  } else {
+    process.exit(exitCode);
+  }
+}
 var logger2 = createLogger("background-worker-cli");
 async function main() {
   const jobId = process.argv[2];
@@ -334,8 +347,9 @@ async function main() {
       );
     }
     logger2.info({ jobId }, "Job completed successfully");
+    await destroyServices(services);
     await shutdownLogger();
-    process.exit(0);
+    forceExitOnMacOS(0);
   } catch (error) {
     logger2.error(
       { jobId, error: error instanceof Error ? error.message : String(error) },
@@ -348,8 +362,9 @@ async function main() {
         "Could not remove PID file after failure"
       );
     }
+    await destroyServices(services);
     await shutdownLogger();
-    process.exit(1);
+    forceExitOnMacOS(1);
   }
 }
 main().catch(async (error) => {
@@ -358,6 +373,6 @@ main().catch(async (error) => {
     "Fatal error in background worker"
   );
   await shutdownLogger();
-  process.exit(1);
+  forceExitOnMacOS(1);
 });
 //# sourceMappingURL=background-worker-cli.js.map

package/dist/workers/background-worker-cli.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../../src/workers/background-worker.ts","../../src/workers/pid-file.ts","../../src/workers/background-worker-cli.ts"],"sourcesContent":["import { createHash } from 'node:crypto';\nimport { IntelligentCrawler, type CrawlProgress } 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  ) {}\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    // 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    // 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();\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      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(),\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        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","#!/usr/bin/env node\nimport { BackgroundWorker } from './background-worker.js';\nimport { writePidFile, deletePidFile, buildPidFilePath } from './pid-file.js';\nimport { createLogger, shutdownLogger } from '../logging/index.js';\nimport { createServices } from '../services/index.js';\nimport { JobService } from '../services/job.service.js';\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  // Create worker and execute job\n  const worker = new BackgroundWorker(\n    jobService,\n    services.store,\n    services.index,\n    services.lance,\n    services.embeddings\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 shutdownLogger();\n    process.exit(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 shutdownLogger();\n    process.exit(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  process.exit(1);\n});\n"],"mappings":";;;;;;;;;;;;;;;AAAA,SAAS,kBAAkB;AAY3B,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,iBACjB;AALiB;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,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,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;AAGvC,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,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,WAAW,oBAAI,KAAK;AAAA,UACtB;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;AAED,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;;;ACjUA,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;;;AC1EA,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,SAAS,IAAI;AAAA,IACjB;AAAA,IACA,SAAS;AAAA,IACT,SAAS;AAAA,IACT,SAAS;AAAA,IACT,SAAS;AAAA,EACX;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,eAAe;AACrB,YAAQ,KAAK,CAAC;AAAA,EAChB,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,eAAe;AACrB,YAAQ,KAAK,CAAC;AAAA,EAChB;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,UAAQ,KAAK,CAAC;AAChB,CAAC;","names":["logger"]}
+{"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  // Create worker and execute job\n  const worker = new BackgroundWorker(\n    jobService,\n    services.store,\n    services.index,\n    services.lance,\n    services.embeddings\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 { IntelligentCrawler, type CrawlProgress } 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  ) {}\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    // 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    // 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();\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      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(),\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        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;AAY3B,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,iBACjB;AALiB;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,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,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;AAGvC,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,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,WAAW,oBAAI,KAAK;AAAA,UACtB;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;AAED,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;;;ACjUA,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,SAAS,IAAI;AAAA,IACjB;AAAA,IACA,SAAS;AAAA,IACT,SAAS;AAAA,IACT,SAAS;AAAA,IACT,SAAS;AAAA,EACX;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

@@ -1,6 +1,6 @@
 {
   "name": "bluera-knowledge",
-  "version": "0.12.7",
+  "version": "0.12.9",
   "description": "CLI tool for managing knowledge stores with semantic search",
   "type": "module",
   "bin": {

package/src/db/embeddings.ts

@@ -64,4 +64,15 @@ export class EmbeddingEngine {
   getDimensions(): number {
     return this.dimensions;
   }
+
+  /**
+   * Dispose the embedding pipeline to free resources.
+   * Should be called before process exit to prevent ONNX runtime cleanup issues on macOS.
+   */
+  async dispose(): Promise<void> {
+    if (this.extractor !== null) {
+      await this.extractor.dispose();
+      this.extractor = null;
+    }
+  }
 }

package/src/db/lance.ts

@@ -1,5 +1,6 @@
 import * as lancedb from '@lancedb/lancedb';
 import { createDocumentId } from '../types/brands.js';
+import { DocumentMetadataSchema } from '../types/document.js';
 import type { StoreId, DocumentId } from '../types/brands.js';
 import type { Document, DocumentMetadata } from '../types/document.js';
 import type { Table, Connection } from '@lancedb/lancedb';
@@ -88,13 +89,17 @@ export class LanceStore {
 
     // Return all results - threshold filtering is applied after score normalization
     // in search.service.ts to match displayed scores
-    return results.map((r) => ({
-      id: createDocumentId(r.id),
-      content: r.content,
-      score: 1 - r._distance,
-      // eslint-disable-next-line @typescript-eslint/consistent-type-assertions
-      metadata: JSON.parse(r.metadata) as DocumentMetadata,
-    }));
+    return results.map((r) => {
+      const metadata = DocumentMetadataSchema.parse(JSON.parse(r.metadata));
+      return {
+        id: createDocumentId(r.id),
+        content: r.content,
+        score: 1 - r._distance,
+        // Schema validates structure, cast to branded type
+        // eslint-disable-next-line @typescript-eslint/consistent-type-assertions
+        metadata: metadata as DocumentMetadata,
+      };
+    });
   }
 
   async createFtsIndex(storeId: StoreId): Promise<void> {
@@ -121,13 +126,17 @@ export class LanceStore {
       _score: number;
     }>;
 
-    return results.map((r) => ({
-      id: createDocumentId(r.id),
-      content: r.content,
-      score: r._score,
-      // eslint-disable-next-line @typescript-eslint/consistent-type-assertions
-      metadata: JSON.parse(r.metadata) as DocumentMetadata,
-    }));
+    return results.map((r) => {
+      const metadata = DocumentMetadataSchema.parse(JSON.parse(r.metadata));
+      return {
+        id: createDocumentId(r.id),
+        content: r.content,
+        score: r._score,
+        // Schema validates structure, cast to branded type
+        // eslint-disable-next-line @typescript-eslint/consistent-type-assertions
+        metadata: metadata as DocumentMetadata,
+      };
+    });
   }
 
   async deleteStore(storeId: StoreId): Promise<void> {

package/src/services/index.ts

@@ -88,6 +88,15 @@ export async function destroyServices(services: ServiceContainer): Promise<void>
     errors.push(error);
   }
 
+  // Dispose embedding engine to free ONNX runtime resources
+  try {
+    await services.embeddings.dispose();
+  } catch (e) {
+    const error = e instanceof Error ? e : new Error(String(e));
+    logger.error({ error }, 'Error disposing EmbeddingEngine');
+    errors.push(error);
+  }
+
   try {
     await services.pythonBridge.stop();
   } catch (e) {

package/src/services/job.service.ts

@@ -1,6 +1,7 @@
 import { randomUUID } from 'crypto';
 import fs from 'fs';
 import path from 'path';
+import { JobSchema } from '../types/job.js';
 import { Result, ok, err } from '../types/result.js';
 import type { Job, CreateJobParams, UpdateJobParams, JobStatus } from '../types/job.js';
 
@@ -90,8 +91,7 @@ export class JobService {
 
     try {
       const content = fs.readFileSync(jobFile, 'utf-8');
-      // eslint-disable-next-line @typescript-eslint/consistent-type-assertions
-      return JSON.parse(content) as Job;
+      return JobSchema.parse(JSON.parse(content));
     } catch (error) {
       throw new Error(
         `Failed to read job ${jobId}: ${error instanceof Error ? error.message : String(error)}`
@@ -117,8 +117,7 @@ export class JobService {
 
       try {
         const content = fs.readFileSync(path.join(this.jobsDir, file), 'utf-8');
-        // eslint-disable-next-line @typescript-eslint/consistent-type-assertions
-        const job = JSON.parse(content) as Job;
+        const job = JobSchema.parse(JSON.parse(content));
 
         if (statusFilter !== undefined) {
           const filters = Array.isArray(statusFilter) ? statusFilter : [statusFilter];

package/src/services/services.test.ts

@@ -1,11 +1,13 @@
 import { describe, it, expect, vi, beforeEach } from 'vitest';
 import { destroyServices, type ServiceContainer } from './index.js';
 import type { PythonBridge } from '../crawl/bridge.js';
+import type { EmbeddingEngine } from '../db/embeddings.js';
 import type { LanceStore } from '../db/lance.js';
 
 describe('destroyServices', () => {
   let mockPythonBridge: { stop: ReturnType<typeof vi.fn> };
   let mockLance: { closeAsync: ReturnType<typeof vi.fn>; close: ReturnType<typeof vi.fn> };
+  let mockEmbeddings: { dispose: ReturnType<typeof vi.fn> };
   let mockServices: ServiceContainer;
 
   beforeEach(() => {
@@ -18,9 +20,14 @@ describe('destroyServices', () => {
       close: vi.fn(),
     };
 
+    mockEmbeddings = {
+      dispose: vi.fn().mockResolvedValue(undefined),
+    };
+
     mockServices = {
       pythonBridge: mockPythonBridge as unknown as PythonBridge,
       lance: mockLance as unknown as LanceStore,
+      embeddings: mockEmbeddings as unknown as EmbeddingEngine,
     } as unknown as ServiceContainer;
   });
 
@@ -84,4 +91,18 @@ describe('destroyServices', () => {
     // Should have waited for closeAsync to complete
     expect(closeCompleted).toBe(true);
   });
+
+  it('calls dispose on EmbeddingEngine to free ONNX runtime resources', async () => {
+    await destroyServices(mockServices);
+
+    expect(mockEmbeddings.dispose).toHaveBeenCalledTimes(1);
+  });
+
+  it('throws on EmbeddingEngine dispose errors', async () => {
+    mockEmbeddings.dispose.mockRejectedValue(new Error('dispose failed'));
+
+    await expect(destroyServices(mockServices)).rejects.toThrow(
+      'Service shutdown failed: dispose failed'
+    );
+  });
 });

package/src/services/store.service.test.ts

@@ -538,6 +538,30 @@ describe('StoreService', () => {
 
       await rm(corruptDir, { recursive: true, force: true });
     });
+
+    it('filters out null entries from stores array on load', async () => {
+      const nullDir = await mkdtemp(join(tmpdir(), 'null-entry-'));
+      const registryPath = join(nullDir, 'stores.json');
+      const validStore = {
+        id: 'test-id',
+        type: 'file',
+        name: 'valid-store',
+        path: '/some/path',
+        status: 'ready',
+        createdAt: new Date().toISOString(),
+        updatedAt: new Date().toISOString(),
+      };
+      await writeFile(registryPath, JSON.stringify({ stores: [null, validStore, null] }));
+
+      const freshService = new StoreService(nullDir);
+      await freshService.initialize();
+
+      const stores = await freshService.list();
+      expect(stores).toHaveLength(1);
+      expect(stores[0]?.name).toBe('valid-store');
+
+      await rm(nullDir, { recursive: true, force: true });
+    });
   });
 
   describe('store definition auto-update', () => {

package/src/services/store.service.ts

@@ -218,6 +218,12 @@ export class StoreService {
           updatedAt: now,
         } satisfies WebStore;
         break;
+
+      default: {
+        // Exhaustive check - if this is reached, input.type is invalid
+        const invalidType: never = input.type;
+        return err(new Error(`Invalid store type: ${String(invalidType)}`));
+      }
     }
 
     this.registry.stores.push(store);
@@ -332,14 +338,16 @@ export class StoreService {
     const content = await readFile(registryPath, 'utf-8');
     try {
       // eslint-disable-next-line @typescript-eslint/consistent-type-assertions
-      const data = JSON.parse(content) as { stores: Store[] };
+      const data = JSON.parse(content) as { stores: (Store | null)[] };
       this.registry = {
-        stores: data.stores.map((s) => ({
-          ...s,
-          id: createStoreId(s.id),
-          createdAt: new Date(s.createdAt),
-          updatedAt: new Date(s.updatedAt),
-        })),
+        stores: data.stores
+          .filter((s): s is Store => s !== null)
+          .map((s) => ({
+            ...s,
+            id: createStoreId(s.id),
+            createdAt: new Date(s.createdAt),
+            updatedAt: new Date(s.updatedAt),
+          })),
       };
     } catch (error) {
       throw new Error(

package/src/types/document.ts

@@ -1,6 +1,30 @@
+import { z } from 'zod';
 import type { DocumentId, StoreId } from './brands.js';
 
-export type DocumentType = 'file' | 'chunk' | 'web';
+// ============================================================================
+// Zod Schemas
+// ============================================================================
+
+export const DocumentTypeSchema = z.enum(['file', 'chunk', 'web']);
+
+export const DocumentMetadataSchema = z
+  .object({
+    path: z.string().optional(),
+    url: z.string().optional(),
+    type: DocumentTypeSchema,
+    storeId: z.string(),
+    indexedAt: z.union([z.string(), z.date()]),
+    fileHash: z.string().optional(),
+    chunkIndex: z.number().optional(),
+    totalChunks: z.number().optional(),
+  })
+  .loose(); // Allow additional fields per index signature
+
+// ============================================================================
+// Types
+// ============================================================================
+
+export type DocumentType = z.infer<typeof DocumentTypeSchema>;
 
 export interface DocumentMetadata {
   readonly path?: string | undefined;

package/src/types/job.ts

@@ -1,36 +1,51 @@
-export type JobType = 'clone' | 'index' | 'crawl';
-export type JobStatus = 'pending' | 'running' | 'completed' | 'failed' | 'cancelled';
-
-export interface JobDetails {
-  storeName?: string;
-  storeId?: string;
-  url?: string;
-  path?: string;
-  filesProcessed?: number;
-  totalFiles?: number;
-  startedAt?: string;
-  completedAt?: string;
-  cancelledAt?: string;
-  error?: string;
+import { z } from 'zod';
+
+// ============================================================================
+// Zod Schemas
+// ============================================================================
+
+export const JobTypeSchema = z.enum(['clone', 'index', 'crawl']);
+export const JobStatusSchema = z.enum(['pending', 'running', 'completed', 'failed', 'cancelled']);
+
+export const JobDetailsSchema = z.object({
+  storeName: z.string().optional(),
+  storeId: z.string().optional(),
+  url: z.string().optional(),
+  path: z.string().optional(),
+  filesProcessed: z.number().optional(),
+  totalFiles: z.number().optional(),
+  startedAt: z.string().optional(),
+  completedAt: z.string().optional(),
+  cancelledAt: z.string().optional(),
+  error: z.string().optional(),
   // Crawl-specific fields
-  crawlInstruction?: string;
-  extractInstruction?: string;
-  maxPages?: number;
-  simple?: boolean;
-  useHeadless?: boolean;
-  pagesCrawled?: number;
-}
+  crawlInstruction: z.string().optional(),
+  extractInstruction: z.string().optional(),
+  maxPages: z.number().optional(),
+  simple: z.boolean().optional(),
+  useHeadless: z.boolean().optional(),
+  pagesCrawled: z.number().optional(),
+});
 
-export interface Job {
-  id: string;
-  type: JobType;
-  status: JobStatus;
-  progress: number; // 0-100
-  message: string;
-  details: JobDetails;
-  createdAt: string;
-  updatedAt: string;
-}
+export const JobSchema = z.object({
+  id: z.string(),
+  type: JobTypeSchema,
+  status: JobStatusSchema,
+  progress: z.number().min(0).max(100),
+  message: z.string(),
+  details: JobDetailsSchema.default({}),
+  createdAt: z.string(),
+  updatedAt: z.string(),
+});
+
+// ============================================================================
+// Types (inferred from schemas)
+// ============================================================================
+
+export type JobType = z.infer<typeof JobTypeSchema>;
+export type JobStatus = z.infer<typeof JobStatusSchema>;
+export type JobDetails = z.infer<typeof JobDetailsSchema>;
+export type Job = z.infer<typeof JobSchema>;
 
 export interface CreateJobParams {
   type: JobType;