bluera-knowledge 0.9.26 → 0.9.31

package/src/services/index.service.ts

@@ -11,6 +11,9 @@ import { ok, err } from '../types/result.js';
 import { ChunkingService } from './chunking.service.js';
 import type { ProgressCallback } from '../types/progress.js';
 import type { CodeGraphService } from './code-graph.service.js';
+import { createLogger } from '../logging/index.js';
+
+const logger = createLogger('index-service');
 
 interface IndexResult {
   documentsIndexed: number;
@@ -51,13 +54,24 @@ export class IndexService {
   }
 
   async indexStore(store: Store, onProgress?: ProgressCallback): Promise<Result<IndexResult>> {
+    logger.info({
+      storeId: store.id,
+      storeName: store.name,
+      storeType: store.type,
+    }, 'Starting store indexing');
+
     try {
       if (store.type === 'file' || store.type === 'repo') {
         return await this.indexFileStore(store, onProgress);
       }
 
+      logger.error({ storeId: store.id, storeType: store.type }, 'Unsupported store type for indexing');
       return err(new Error(`Indexing not supported for store type: ${store.type}`));
     } catch (error) {
+      logger.error({
+        storeId: store.id,
+        error: error instanceof Error ? error.message : String(error),
+      }, 'Store indexing failed');
       return err(error instanceof Error ? error : new Error(String(error)));
     }
   }
@@ -68,6 +82,12 @@ export class IndexService {
     const documents: Document[] = [];
     let filesProcessed = 0;
 
+    logger.debug({
+      storeId: store.id,
+      path: store.path,
+      fileCount: files.length,
+    }, 'Files scanned for indexing');
+
     // Collect source files for code graph building
     const sourceFiles: Array<{ path: string; content: string }> = [];
 
@@ -152,10 +172,21 @@ export class IndexService {
       message: 'Indexing complete'
     });
 
+    const timeMs = Date.now() - startTime;
+
+    logger.info({
+      storeId: store.id,
+      storeName: store.name,
+      documentsIndexed: filesProcessed,
+      chunksCreated: documents.length,
+      sourceFilesForGraph: sourceFiles.length,
+      timeMs,
+    }, 'Store indexing complete');
+
     return ok({
       documentsIndexed: filesProcessed,
       chunksCreated: documents.length,
-      timeMs: Date.now() - startTime,
+      timeMs,
     });
   }
 
@@ -264,3 +295,47 @@ export class IndexService {
     return false;
   }
 }
+
+/**
+ * Classify web content type based on URL patterns and page title.
+ * Used for ranking boosts similar to local file classification.
+ */
+export function classifyWebContentType(url: string, title?: string): string {
+  const urlLower = url.toLowerCase();
+  const titleLower = (title ?? '').toLowerCase();
+
+  // API reference documentation → documentation-primary (1.8x boost)
+  if (/\/api[-/]?(ref|reference|docs?)?\//i.test(urlLower) ||
+      /api\s*(reference|documentation)/i.test(titleLower)) {
+    return 'documentation-primary';
+  }
+
+  // Getting started / tutorials → documentation-primary (1.8x boost)
+  if (/\/(getting[-_]?started|quickstart|tutorial|setup)\b/i.test(urlLower) ||
+      /(getting started|quickstart|tutorial)/i.test(titleLower)) {
+    return 'documentation-primary';
+  }
+
+  // General docs paths → documentation (1.5x boost)
+  if (/\/(docs?|documentation|reference|learn|manual|guide)/i.test(urlLower)) {
+    return 'documentation';
+  }
+
+  // Examples and demos → example (1.4x boost)
+  if (/\/(examples?|demos?|samples?|cookbook)/i.test(urlLower)) {
+    return 'example';
+  }
+
+  // Changelog → changelog (special handling in intent boosts)
+  if (/changelog|release[-_]?notes/i.test(urlLower)) {
+    return 'changelog';
+  }
+
+  // Blog posts → lower priority
+  if (/\/blog\//i.test(urlLower)) {
+    return 'other';
+  }
+
+  // Web content without specific path indicators is treated as documentation
+  return 'documentation';
+}

package/src/services/index.ts

@@ -6,6 +6,9 @@ import { CodeGraphService } from './code-graph.service.js';
 import { LanceStore } from '../db/lance.js';
 import { EmbeddingEngine } from '../db/embeddings.js';
 import { PythonBridge } from '../crawl/bridge.js';
+import { createLogger, shutdownLogger } from '../logging/index.js';
+
+const logger = createLogger('services');
 
 export { ConfigService } from './config.service.js';
 export { StoreService } from './store.service.js';
@@ -32,6 +35,8 @@ export async function createServices(
   dataDir?: string,
   projectRoot?: string
 ): Promise<ServiceContainer> {
+  logger.info({ configPath, dataDir, projectRoot }, 'Initializing services');
+
   const config = new ConfigService(configPath, dataDir, projectRoot);
   const appConfig = await config.load();
   const resolvedDataDir = config.resolveDataDir();
@@ -51,9 +56,11 @@ export async function createServices(
   await pythonBridge.start();
 
   const codeGraph = new CodeGraphService(resolvedDataDir, pythonBridge);
-  const search = new SearchService(lance, embeddings, undefined, codeGraph);
+  const search = new SearchService(lance, embeddings, codeGraph);
   const index = new IndexService(lance, embeddings, { codeGraphService: codeGraph });
 
+  logger.info({ dataDir: resolvedDataDir }, 'Services initialized successfully');
+
   return {
     config,
     store,
@@ -71,5 +78,16 @@ export async function createServices(
  * Call this after CLI commands complete to allow the process to exit.
  */
 export async function destroyServices(services: ServiceContainer): Promise<void> {
-  await services.pythonBridge.stop();
+  logger.info('Shutting down services');
+  try {
+    services.lance.close();
+  } catch (e) {
+    logger.error({ error: e }, 'Error closing LanceStore');
+  }
+  try {
+    await services.pythonBridge.stop();
+  } catch (e) {
+    logger.error({ error: e }, 'Error stopping Python bridge');
+  }
+  await shutdownLogger();
 }