feat: add configurable embedding batch size setting

- Added new VSCode setting roo-cline.codeIndex.embeddingBatchSize
- Default value is 400, configurable range 1-2048
- Updated DirectoryScanner and FileWatcher to use configurable batch size
- Updated service factory to pass batch size to processors
- Maintains backward compatibility with default value
- Fixes #7356 - allows users to configure batch size based on API provider limits

Author: roomote

src/package.json

@@ -400,6 +400,13 @@
 					"type": "boolean",
 					"default": false,
 					"description": "%settings.newTaskRequireTodos.description%"
+				},
+				"roo-cline.codeIndex.embeddingBatchSize": {
+					"type": "number",
+					"default": 400,
+					"minimum": 1,
+					"maximum": 2048,
+					"description": "The batch size for embedding operations during code indexing. Adjust this based on your API provider's limits. Default is 400."
 				}
 			}
 		}

src/services/code-index/processors/file-watcher.ts

@@ -26,6 +26,7 @@ import { isPathInIgnoredDirectory } from "../../glob/ignore-utils"
 import { TelemetryService } from "@roo-code/telemetry"
 import { TelemetryEventName } from "@roo-code/types"
 import { sanitizeErrorMessage } from "../shared/validation-helpers"
+import { Package } from "../../../shared/package"
 
 /**
  * Implementation of the file watcher interface
@@ -38,6 +39,7 @@ export class FileWatcher implements IFileWatcher {
 	private batchProcessDebounceTimer?: NodeJS.Timeout
 	private readonly BATCH_DEBOUNCE_DELAY_MS = 500
 	private readonly FILE_PROCESSING_CONCURRENCY_LIMIT = 10
+	private readonly batchSegmentThreshold: number
 
 	private readonly _onDidStartBatchProcessing = new vscode.EventEmitter<string[]>()
 	private readonly _onBatchProgressUpdate = new vscode.EventEmitter<{
@@ -78,11 +80,26 @@ export class FileWatcher implements IFileWatcher {
 		private vectorStore?: IVectorStore,
 		ignoreInstance?: Ignore,
 		ignoreController?: RooIgnoreController,
+		batchSegmentThreshold?: number,
 	) {
 		this.ignoreController = ignoreController || new RooIgnoreController(workspacePath)
 		if (ignoreInstance) {
 			this.ignoreInstance = ignoreInstance
 		}
+		// Get the configurable batch size from VSCode settings, fallback to default
+		// If not provided in constructor, try to get from VSCode settings
+		if (batchSegmentThreshold !== undefined) {
+			this.batchSegmentThreshold = batchSegmentThreshold
+		} else {
+			try {
+				this.batchSegmentThreshold = vscode.workspace
+					.getConfiguration(Package.name)
+					.get<number>("codeIndex.embeddingBatchSize", BATCH_SEGMENT_THRESHOLD)
+			} catch {
+				// In test environment, vscode.workspace might not be available
+				this.batchSegmentThreshold = BATCH_SEGMENT_THRESHOLD
+			}
+		}
 	}
 
 	/**
@@ -341,8 +358,8 @@ export class FileWatcher implements IFileWatcher {
 	): Promise<Error | undefined> {
 		if (pointsForBatchUpsert.length > 0 && this.vectorStore && !overallBatchError) {
 			try {
-				for (let i = 0; i < pointsForBatchUpsert.length; i += BATCH_SEGMENT_THRESHOLD) {
-					const batch = pointsForBatchUpsert.slice(i, i + BATCH_SEGMENT_THRESHOLD)
+				for (let i = 0; i < pointsForBatchUpsert.length; i += this.batchSegmentThreshold) {
+					const batch = pointsForBatchUpsert.slice(i, i + this.batchSegmentThreshold)
 					let retryCount = 0
 					let upsertError: Error | undefined
 

src/services/code-index/processors/scanner.ts

@@ -29,15 +29,34 @@ import { isPathInIgnoredDirectory } from "../../glob/ignore-utils"
 import { TelemetryService } from "@roo-code/telemetry"
 import { TelemetryEventName } from "@roo-code/types"
 import { sanitizeErrorMessage } from "../shared/validation-helpers"
+import { Package } from "../../../shared/package"
 
 export class DirectoryScanner implements IDirectoryScanner {
+	private readonly batchSegmentThreshold: number
+
 	constructor(
 		private readonly embedder: IEmbedder,
 		private readonly qdrantClient: IVectorStore,
 		private readonly codeParser: ICodeParser,
 		private readonly cacheManager: CacheManager,
 		private readonly ignoreInstance: Ignore,
-	) {}
+		batchSegmentThreshold?: number,
+	) {
+		// Get the configurable batch size from VSCode settings, fallback to default
+		// If not provided in constructor, try to get from VSCode settings
+		if (batchSegmentThreshold !== undefined) {
+			this.batchSegmentThreshold = batchSegmentThreshold
+		} else {
+			try {
+				this.batchSegmentThreshold = vscode.workspace
+					.getConfiguration(Package.name)
+					.get<number>("codeIndex.embeddingBatchSize", BATCH_SEGMENT_THRESHOLD)
+			} catch {
+				// In test environment, vscode.workspace might not be available
+				this.batchSegmentThreshold = BATCH_SEGMENT_THRESHOLD
+			}
+		}
+	}
 
 	/**
 	 * Recursively scans a directory for code blocks in supported files.
@@ -153,7 +172,7 @@ export class DirectoryScanner implements IDirectoryScanner {
 									addedBlocksFromFile = true
 
 									// Check if batch threshold is met
-									if (currentBatchBlocks.length >= BATCH_SEGMENT_THRESHOLD) {
+									if (currentBatchBlocks.length >= this.batchSegmentThreshold) {
 										// Wait if we've reached the maximum pending batches
 										while (pendingBatchCount >= MAX_PENDING_BATCHES) {
 											// Wait for at least one batch to complete

src/services/code-index/service-factory.ts

@@ -16,6 +16,8 @@ import { Ignore } from "ignore"
 import { t } from "../../i18n"
 import { TelemetryService } from "@roo-code/telemetry"
 import { TelemetryEventName } from "@roo-code/types"
+import { Package } from "../../shared/package"
+import { BATCH_SEGMENT_THRESHOLD } from "./constants"
 
 /**
  * Factory class responsible for creating and configuring code indexing service dependencies.
@@ -156,7 +158,17 @@ export class CodeIndexServiceFactory {
 		parser: ICodeParser,
 		ignoreInstance: Ignore,
 	): DirectoryScanner {
-		return new DirectoryScanner(embedder, vectorStore, parser, this.cacheManager, ignoreInstance)
+		// Get the configurable batch size from VSCode settings
+		let batchSize: number | undefined
+		try {
+			batchSize = vscode.workspace
+				.getConfiguration(Package.name)
+				.get<number>("codeIndex.embeddingBatchSize", BATCH_SEGMENT_THRESHOLD)
+		} catch {
+			// In test environment, vscode.workspace might not be available
+			batchSize = undefined
+		}
+		return new DirectoryScanner(embedder, vectorStore, parser, this.cacheManager, ignoreInstance, batchSize)
 	}
 
 	/**
@@ -170,6 +182,16 @@ export class CodeIndexServiceFactory {
 		ignoreInstance: Ignore,
 		rooIgnoreController?: RooIgnoreController,
 	): IFileWatcher {
+		// Get the configurable batch size from VSCode settings
+		let batchSize: number | undefined
+		try {
+			batchSize = vscode.workspace
+				.getConfiguration(Package.name)
+				.get<number>("codeIndex.embeddingBatchSize", BATCH_SEGMENT_THRESHOLD)
+		} catch {
+			// In test environment, vscode.workspace might not be available
+			batchSize = undefined
+		}
 		return new FileWatcher(
 			this.workspacePath,
 			context,
@@ -178,6 +200,7 @@ export class CodeIndexServiceFactory {
 			vectorStore,
 			ignoreInstance,
 			rooIgnoreController,
+			batchSize,
 		)
 	}