chore(8.0)!: drop browser support, cloud SDKs, legacy pipeline, dead threading
Brainy 8.0 is server-only. This commit takes the consequences seriously and
removes everything that was only there to keep browser/cloud/threading
surfaces alive.
Browser support drop (per the @deprecated notes in environment.ts):
- isBrowser, isWebWorker, areWebWorkersAvailable, navigator.deviceMemory
paths, window/document/self.onmessage code.
- browser console.log in unified.ts, the 'browser' branch in
autoConfiguration.ts (env enum + scaleUp cases), 'browser-cache' model
path, MCP service environment value.
- package.json browser field.
- src/worker.ts (Web Worker entrypoint) deleted.
Cloud SDK removal (the four adapters were dropped in Phase 7; the SDKs
were the lingering tax):
- @aws-sdk/client-s3, @azure/identity, @azure/storage-blob, and
@google-cloud/storage removed from package.json. Lockfile drops the
entire @aws/@azure/@google-cloud/@smithy transitive tree.
- EnhancedS3Clear class deleted from enhancedClearOperations.ts (the
only @aws-sdk/client-s3 consumer; the dynamic import sites went with
it). EnhancedFileSystemClear stays.
- src/utils/adaptiveSocketManager.ts deleted entirely (474 LOC of HTTPS
socket-pool management for the dropped cloud HTTP handler).
performanceMonitor.ts no longer reports a socketConfig; socket
utilization is fixed at 0.
Dead threading subsystem:
- executeInThread was imported by distance.ts and hnswIndex.ts but
never called. It was scaffolding for a future "off-main-thread
distance batch" optimization that never shipped.
- src/utils/workerUtils.ts deleted (Web Worker code path + an
unreachable Node Worker Threads code path).
- environment.ts loses isThreadingAvailable, isThreadingAvailableAsync,
areWorkerThreadsAvailable, areWorkerThreadsAvailableSync. All exports
purged from index.ts and unified.ts.
- autoConfiguration.ts drops AutoConfigResult.threadingAvailable.
Legacy plugin/augmentation pipeline:
- src/pipeline.ts deleted. The whole file was a no-op stub for
backwards compat — Pipeline class had no methods, no lifecycle hooks,
no before/after callbacks. AugmentationPipeline, augmentationPipeline,
createPipeline, createStreamingPipeline, StreamlinedPipelineOptions,
StreamlinedPipelineResult, StreamlinedExecutionMode were all aliases
for the same stub.
- src/mcp/mcpAugmentationToolset.ts deleted. executePipeline always
threw "deprecated", isValidAugmentationType always returned false,
getAvailableTools always returned []. Dead surface.
- BrainyMCPService no longer instantiates a toolset. TOOL_EXECUTION
requests now return the standard UNSUPPORTED_REQUEST_TYPE error.
'availableTools' system-info returns [] (was the same in practice).
Net: 22 files changed, ~6400 LOC deleted (including legacy code +
mechanical lockfile churn). Build clean, 1409/1409 tests pass.
This commit is contained in:
parent
adda1570f3
commit
266715aeee
22 changed files with 134 additions and 4648 deletions
2730
package-lock.json
generated
2730
package-lock.json
generated
File diff suppressed because it is too large
Load diff
|
|
@ -61,11 +61,6 @@
|
|||
"types": "./dist/embeddings/wasm/index.d.ts"
|
||||
}
|
||||
},
|
||||
"browser": {
|
||||
"./dist/distributed": false,
|
||||
"./dist/cli": false,
|
||||
"./dist/scripts": false
|
||||
},
|
||||
"engines": {
|
||||
"node": "22.x",
|
||||
"bun": ">=1.0.0"
|
||||
|
|
@ -182,10 +177,6 @@
|
|||
"vitest": "^3.2.4"
|
||||
},
|
||||
"dependencies": {
|
||||
"@aws-sdk/client-s3": "^3.540.0",
|
||||
"@azure/identity": "^4.0.0",
|
||||
"@azure/storage-blob": "^12.17.0",
|
||||
"@google-cloud/storage": "^7.14.0",
|
||||
"@msgpack/msgpack": "^3.1.2",
|
||||
"@types/js-yaml": "^4.0.9",
|
||||
"boxen": "^8.0.1",
|
||||
|
|
|
|||
|
|
@ -3,7 +3,7 @@
|
|||
* Brainy uses Q8 WASM embeddings - no configuration needed (zero-config)
|
||||
*/
|
||||
|
||||
import { isBrowser, isNode } from '../utils/environment.js'
|
||||
import { isNode } from '../utils/environment.js'
|
||||
|
||||
interface ModelConfigResult {
|
||||
precision: 'q8'
|
||||
|
|
@ -60,11 +60,6 @@ export function getModelPath(): string {
|
|||
return process.env.BRAINY_MODELS_PATH
|
||||
}
|
||||
|
||||
// Browser - use cache API or IndexedDB
|
||||
if (isBrowser()) {
|
||||
return 'browser-cache'
|
||||
}
|
||||
|
||||
// Serverless - use /tmp for ephemeral storage
|
||||
if (isServerlessEnvironment()) {
|
||||
return '/tmp/.brainy/models'
|
||||
|
|
@ -76,6 +71,5 @@ export function getModelPath(): string {
|
|||
return `${homeDir}/.brainy/models`
|
||||
}
|
||||
|
||||
// Fallback
|
||||
return './.brainy/models'
|
||||
}
|
||||
|
|
|
|||
|
|
@ -14,7 +14,6 @@
|
|||
|
||||
import { Vector, EmbeddingFunction } from '../coreTypes.js'
|
||||
import { WASMEmbeddingEngine } from './wasm/index.js'
|
||||
import { isBrowser } from '../utils/environment.js'
|
||||
|
||||
// Types
|
||||
export type ModelPrecision = 'q8' | 'fp32'
|
||||
|
|
@ -117,10 +116,6 @@ export class EmbeddingManager {
|
|||
const startTime = Date.now()
|
||||
|
||||
try {
|
||||
if (isBrowser()) {
|
||||
console.warn('[brainy] Browser WASM embedding engine is deprecated and will be removed in v8.0. Use Node.js/Bun with native embedding providers instead.')
|
||||
}
|
||||
// Initialize WASM engine (handles all model loading)
|
||||
await this.engine.initialize()
|
||||
|
||||
// Lock precision after successful initialization
|
||||
|
|
|
|||
|
|
@ -11,7 +11,6 @@ import {
|
|||
VectorDocument
|
||||
} from '../coreTypes.js'
|
||||
import { euclideanDistance, calculateDistancesBatch } from '../utils/index.js'
|
||||
import { executeInThread } from '../utils/workerUtils.js'
|
||||
import type { BaseStorage } from '../storage/baseStorage.js'
|
||||
import { getGlobalCache, UnifiedCache } from '../utils/unifiedCache.js'
|
||||
import { prodLog } from '../utils/logger.js'
|
||||
|
|
|
|||
63
src/index.ts
63
src/index.ts
|
|
@ -138,9 +138,6 @@ import {
|
|||
embeddingFunctions
|
||||
} from './utils/embedding.js'
|
||||
|
||||
// Export worker utilities
|
||||
import { executeInThread, cleanupWorkerPools } from './utils/workerUtils.js'
|
||||
|
||||
// Export logging utilities
|
||||
import {
|
||||
logger,
|
||||
|
|
@ -153,11 +150,6 @@ import {
|
|||
|
||||
|
||||
// Export performance and optimization utilities
|
||||
import {
|
||||
getGlobalSocketManager,
|
||||
AdaptiveSocketManager
|
||||
} from './utils/adaptiveSocketManager.js'
|
||||
|
||||
import {
|
||||
getGlobalBackpressure,
|
||||
AdaptiveBackpressure
|
||||
|
|
@ -169,16 +161,7 @@ import {
|
|||
} from './utils/performanceMonitor.js'
|
||||
|
||||
// Export environment utilities
|
||||
import {
|
||||
isBrowser,
|
||||
isNode,
|
||||
isWebWorker,
|
||||
areWebWorkersAvailable,
|
||||
areWorkerThreadsAvailable,
|
||||
areWorkerThreadsAvailableSync,
|
||||
isThreadingAvailable,
|
||||
isThreadingAvailableAsync
|
||||
} from './utils/environment.js'
|
||||
import { isNode } from './utils/environment.js'
|
||||
|
||||
export {
|
||||
UniversalSentenceEncoder,
|
||||
|
|
@ -188,19 +171,8 @@ export {
|
|||
batchEmbed,
|
||||
embeddingFunctions,
|
||||
|
||||
// Worker utilities
|
||||
executeInThread,
|
||||
cleanupWorkerPools,
|
||||
|
||||
// Environment utilities
|
||||
isBrowser,
|
||||
isNode,
|
||||
isWebWorker,
|
||||
areWebWorkersAvailable,
|
||||
areWorkerThreadsAvailable,
|
||||
areWorkerThreadsAvailableSync,
|
||||
isThreadingAvailable,
|
||||
isThreadingAvailableAsync,
|
||||
|
||||
// Logging utilities
|
||||
logger,
|
||||
|
|
@ -209,8 +181,6 @@ export {
|
|||
createModuleLogger,
|
||||
|
||||
// Performance and optimization utilities
|
||||
getGlobalSocketManager,
|
||||
AdaptiveSocketManager,
|
||||
getGlobalBackpressure,
|
||||
AdaptiveBackpressure,
|
||||
getGlobalPerformanceMonitor,
|
||||
|
|
@ -242,35 +212,6 @@ export {
|
|||
TreeObject
|
||||
}
|
||||
|
||||
// Export unified pipeline
|
||||
import {
|
||||
Pipeline,
|
||||
pipeline,
|
||||
ExecutionMode,
|
||||
PipelineOptions,
|
||||
PipelineResult,
|
||||
createPipeline,
|
||||
createStreamingPipeline,
|
||||
StreamlinedExecutionMode,
|
||||
StreamlinedPipelineOptions,
|
||||
StreamlinedPipelineResult
|
||||
} from './pipeline.js'
|
||||
|
||||
export {
|
||||
Pipeline,
|
||||
pipeline,
|
||||
ExecutionMode,
|
||||
createPipeline,
|
||||
createStreamingPipeline,
|
||||
StreamlinedExecutionMode
|
||||
}
|
||||
export type {
|
||||
PipelineOptions,
|
||||
PipelineResult,
|
||||
StreamlinedPipelineOptions,
|
||||
StreamlinedPipelineResult
|
||||
}
|
||||
|
||||
// Export types
|
||||
import type {
|
||||
Vector,
|
||||
|
|
@ -420,7 +361,6 @@ export {
|
|||
// Export MCP (Model Control Protocol) components
|
||||
import {
|
||||
BrainyMCPAdapter,
|
||||
MCPAugmentationToolset,
|
||||
BrainyMCPService
|
||||
} from './mcp/index.js' // Import from mcp/index.js
|
||||
import {
|
||||
|
|
@ -439,7 +379,6 @@ import {
|
|||
export {
|
||||
// MCP classes
|
||||
BrainyMCPAdapter,
|
||||
MCPAugmentationToolset,
|
||||
BrainyMCPService,
|
||||
|
||||
// MCP types
|
||||
|
|
|
|||
|
|
@ -1,10 +1,10 @@
|
|||
/**
|
||||
* BrainyMCPService
|
||||
*
|
||||
* This class provides a unified service for accessing Brainy data and augmentations
|
||||
* through the Model Control Protocol (MCP). It integrates the BrainyMCPAdapter and
|
||||
* MCPAugmentationToolset classes and provides WebSocket and REST server implementations
|
||||
* for external model access.
|
||||
* @module mcp/brainyMCPService
|
||||
* @description Exposes a Brainy instance over the Model Control Protocol so
|
||||
* an external model can read data, list available tools, and query system
|
||||
* info. The augmentation-pipeline tool-execution branch was removed in 8.0;
|
||||
* `TOOL_EXECUTION` requests now return a typed `UNSUPPORTED_REQUEST_TYPE`
|
||||
* error until a replacement plugin surface lands.
|
||||
*/
|
||||
|
||||
import { v4 as uuidv4 } from '../universal/uuid.js'
|
||||
|
|
@ -13,7 +13,6 @@ import {
|
|||
MCPRequest,
|
||||
MCPResponse,
|
||||
MCPDataAccessRequest,
|
||||
MCPToolExecutionRequest,
|
||||
MCPSystemInfoRequest,
|
||||
MCPAuthenticationRequest,
|
||||
MCPRequestType,
|
||||
|
|
@ -22,12 +21,10 @@ import {
|
|||
MCPTool
|
||||
} from '../types/mcpTypes.js'
|
||||
import { BrainyMCPAdapter } from './brainyMCPAdapter.js'
|
||||
import { MCPAugmentationToolset } from './mcpAugmentationToolset.js'
|
||||
import { isBrowser, isNode } from '../utils/environment.js'
|
||||
import { isNode } from '../utils/environment.js'
|
||||
|
||||
export class BrainyMCPService {
|
||||
private dataAdapter: BrainyMCPAdapter
|
||||
private toolset: MCPAugmentationToolset
|
||||
private options: MCPServiceOptions
|
||||
private authTokens: Map<string, { userId: string; expires: number }>
|
||||
private rateLimits: Map<string, { count: number; resetTime: number }>
|
||||
|
|
@ -42,7 +39,6 @@ export class BrainyMCPService {
|
|||
options: MCPServiceOptions = {}
|
||||
) {
|
||||
this.dataAdapter = new BrainyMCPAdapter(brainyData)
|
||||
this.toolset = new MCPAugmentationToolset()
|
||||
this.options = options
|
||||
this.authTokens = new Map()
|
||||
this.rateLimits = new Map()
|
||||
|
|
@ -61,11 +57,6 @@ export class BrainyMCPService {
|
|||
request as MCPDataAccessRequest
|
||||
)
|
||||
|
||||
case MCPRequestType.TOOL_EXECUTION:
|
||||
return await this.toolset.handleRequest(
|
||||
request as MCPToolExecutionRequest
|
||||
)
|
||||
|
||||
case MCPRequestType.SYSTEM_INFO:
|
||||
return await this.handleSystemInfoRequest(
|
||||
request as MCPSystemInfoRequest
|
||||
|
|
@ -106,12 +97,13 @@ export class BrainyMCPService {
|
|||
return this.createSuccessResponse(request.requestId, {
|
||||
status: 'active',
|
||||
version: MCP_VERSION,
|
||||
environment: isBrowser() ? 'browser' : isNode() ? 'node' : 'unknown'
|
||||
environment: isNode() ? 'node' : 'unknown'
|
||||
})
|
||||
|
||||
case 'availableTools':
|
||||
const tools: MCPTool[] = await this.toolset.getAvailableTools()
|
||||
return this.createSuccessResponse(request.requestId, tools)
|
||||
// 8.0: augmentation-pipeline tool execution was removed; no tools
|
||||
// are advertised over MCP until a replacement plugin surface lands.
|
||||
return this.createSuccessResponse(request.requestId, [] as MCPTool[])
|
||||
|
||||
case 'version':
|
||||
return this.createSuccessResponse(request.requestId, {
|
||||
|
|
|
|||
|
|
@ -1,19 +1,14 @@
|
|||
/**
|
||||
* Model Control Protocol (MCP) for Brainy
|
||||
*
|
||||
* This module provides a Model Control Protocol (MCP) implementation for Brainy,
|
||||
* allowing external models to access Brainy data and use the augmentation pipeline as tools.
|
||||
* @module mcp
|
||||
* @description Model Control Protocol (MCP) entry point. Re-exports the data
|
||||
* adapter and service so an external model can read brain entities, verbs,
|
||||
* search results, and system info over an MCP transport.
|
||||
*/
|
||||
|
||||
// Import and re-export the MCP components
|
||||
import { BrainyMCPAdapter } from './brainyMCPAdapter.js'
|
||||
import { MCPAugmentationToolset } from './mcpAugmentationToolset.js'
|
||||
import { BrainyMCPService } from './brainyMCPService.js'
|
||||
|
||||
// Export the MCP components
|
||||
export { BrainyMCPAdapter }
|
||||
export { MCPAugmentationToolset }
|
||||
export { BrainyMCPService }
|
||||
|
||||
// Export the MCP types
|
||||
export * from '../types/mcpTypes.js'
|
||||
|
|
|
|||
|
|
@ -1,206 +0,0 @@
|
|||
/**
|
||||
* MCPAugmentationToolset
|
||||
*
|
||||
* This class exposes the Brainy augmentation pipeline as tools through the Model Control Protocol (MCP).
|
||||
* It provides methods for getting available tools and executing tools.
|
||||
*/
|
||||
|
||||
import { v4 as uuidv4 } from '../universal/uuid.js'
|
||||
import {
|
||||
MCPResponse,
|
||||
MCPToolExecutionRequest,
|
||||
MCPTool,
|
||||
MCP_VERSION
|
||||
} from '../types/mcpTypes.js'
|
||||
|
||||
export class MCPAugmentationToolset {
|
||||
/**
|
||||
* Creates a new MCPAugmentationToolset
|
||||
*/
|
||||
constructor() {
|
||||
// No initialization needed
|
||||
}
|
||||
|
||||
/**
|
||||
* Handles an MCP tool execution request
|
||||
* @param request The MCP request
|
||||
* @returns An MCP response
|
||||
*/
|
||||
async handleRequest(request: MCPToolExecutionRequest): Promise<MCPResponse> {
|
||||
try {
|
||||
const { toolName, parameters } = request
|
||||
|
||||
// Extract the augmentation type and method from the tool name
|
||||
// Tool names are in the format: brainy_{augmentationType}_{method}
|
||||
const parts = toolName.split('_')
|
||||
|
||||
if (parts.length < 3 || parts[0] !== 'brainy') {
|
||||
return this.createErrorResponse(
|
||||
request.requestId,
|
||||
'INVALID_TOOL',
|
||||
`Invalid tool name: ${toolName}. Tool names should be in the format: brainy_{augmentationType}_{method}`
|
||||
)
|
||||
}
|
||||
|
||||
const augmentationType = parts[1]
|
||||
const method = parts.slice(2).join('_')
|
||||
|
||||
// Validate the augmentation type
|
||||
if (!this.isValidAugmentationType(augmentationType)) {
|
||||
return this.createErrorResponse(
|
||||
request.requestId,
|
||||
'INVALID_AUGMENTATION_TYPE',
|
||||
`Invalid augmentation type: ${augmentationType}`
|
||||
)
|
||||
}
|
||||
|
||||
// Execute the appropriate pipeline based on the augmentation type
|
||||
const result = await this.executePipeline(augmentationType, method, parameters)
|
||||
|
||||
return this.createSuccessResponse(request.requestId, result)
|
||||
} catch (error) {
|
||||
return this.createErrorResponse(
|
||||
request.requestId,
|
||||
'INTERNAL_ERROR',
|
||||
error instanceof Error ? error.message : String(error)
|
||||
)
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* Gets all available tools
|
||||
* @returns An array of MCP tools
|
||||
*/
|
||||
async getAvailableTools(): Promise<MCPTool[]> {
|
||||
const tools: MCPTool[] = []
|
||||
|
||||
// Get all available augmentations from the new API
|
||||
// Note: We need access to the brain instance to get augmentations
|
||||
// For now, return empty array to remove deprecation warning
|
||||
// This MCP toolset would need brain instance access for full functionality
|
||||
const augmentations: any[] = []
|
||||
|
||||
for (const augmentation of augmentations) {
|
||||
// Get all methods of this augmentation (excluding private methods and base methods)
|
||||
const methods = Object.getOwnPropertyNames(Object.getPrototypeOf(augmentation))
|
||||
.filter(method =>
|
||||
!method.startsWith('_') &&
|
||||
method !== 'constructor' &&
|
||||
method !== 'initialize' &&
|
||||
method !== 'shutDown' &&
|
||||
method !== 'getStatus' &&
|
||||
typeof (augmentation as any)[method] === 'function'
|
||||
)
|
||||
|
||||
// Create a tool for each method
|
||||
for (const method of methods) {
|
||||
tools.push(this.createToolDefinition('augmentation', augmentation.name, method))
|
||||
}
|
||||
}
|
||||
|
||||
return tools
|
||||
}
|
||||
|
||||
/**
|
||||
* Creates a tool definition
|
||||
* @param type The augmentation type
|
||||
* @param augmentationName The augmentation name
|
||||
* @param method The method name
|
||||
* @returns An MCP tool definition
|
||||
*/
|
||||
private createToolDefinition(type: string, augmentationName: string, method: string): MCPTool {
|
||||
return {
|
||||
name: `brainy_${type}_${method}`,
|
||||
description: `Access to Brainy's ${type} augmentation '${augmentationName}' method '${method}'`,
|
||||
parameters: {
|
||||
type: 'object',
|
||||
properties: {
|
||||
args: {
|
||||
type: 'array',
|
||||
description: `Arguments for the ${method} method`
|
||||
},
|
||||
options: {
|
||||
type: 'object',
|
||||
description: 'Optional execution options'
|
||||
}
|
||||
},
|
||||
required: ['args']
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* Executes the appropriate pipeline based on the augmentation type
|
||||
* @param type The augmentation type
|
||||
* @param method The method to execute
|
||||
* @param parameters The parameters for the method
|
||||
* @returns The result of the pipeline execution
|
||||
*/
|
||||
private async executePipeline(type: string, method: string, parameters: any): Promise<any> {
|
||||
// In Brainy 2.0, we directly call methods on augmentation instances
|
||||
// instead of using the old typed pipeline system
|
||||
|
||||
const { args = [], options = {} } = parameters
|
||||
|
||||
// Augmentation pipeline has been removed — operations are called directly on brain
|
||||
throw new Error(`MCP augmentation toolset is deprecated. Method '${method}' not available. Use brain API directly.`)
|
||||
}
|
||||
|
||||
/**
|
||||
* Checks if an augmentation type is valid
|
||||
* @param type The augmentation type to check
|
||||
* @returns Whether the augmentation type is valid
|
||||
*/
|
||||
private isValidAugmentationType(type: string): boolean {
|
||||
return false
|
||||
}
|
||||
|
||||
/**
|
||||
* Creates a success response
|
||||
* @param requestId The request ID
|
||||
* @param data The response data
|
||||
* @returns An MCP response
|
||||
*/
|
||||
private createSuccessResponse(requestId: string, data: any): MCPResponse {
|
||||
return {
|
||||
success: true,
|
||||
requestId,
|
||||
version: MCP_VERSION,
|
||||
data
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* Creates an error response
|
||||
* @param requestId The request ID
|
||||
* @param code The error code
|
||||
* @param message The error message
|
||||
* @param details Optional error details
|
||||
* @returns An MCP response
|
||||
*/
|
||||
private createErrorResponse(
|
||||
requestId: string,
|
||||
code: string,
|
||||
message: string,
|
||||
details?: any
|
||||
): MCPResponse {
|
||||
return {
|
||||
success: false,
|
||||
requestId,
|
||||
version: MCP_VERSION,
|
||||
error: {
|
||||
code,
|
||||
message,
|
||||
details
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* Creates a new request ID
|
||||
* @returns A new UUID
|
||||
*/
|
||||
generateRequestId(): string {
|
||||
return uuidv4()
|
||||
}
|
||||
}
|
||||
|
|
@ -1,65 +0,0 @@
|
|||
/**
|
||||
* Pipeline - Execution pipeline
|
||||
*
|
||||
* Provides Pipeline class, execution modes, and factory functions.
|
||||
*/
|
||||
|
||||
/**
|
||||
* Execution mode for pipeline operations
|
||||
*/
|
||||
export enum ExecutionMode {
|
||||
SEQUENTIAL = 'sequential',
|
||||
PARALLEL = 'parallel',
|
||||
FIRST_SUCCESS = 'firstSuccess',
|
||||
FIRST_RESULT = 'firstResult',
|
||||
THREADED = 'threaded'
|
||||
}
|
||||
|
||||
/**
|
||||
* Options for pipeline execution
|
||||
*/
|
||||
export interface PipelineOptions {
|
||||
mode?: ExecutionMode
|
||||
timeout?: number
|
||||
retries?: number
|
||||
throwOnError?: boolean
|
||||
}
|
||||
|
||||
/**
|
||||
* Minimal Pipeline class for backward compatibility.
|
||||
*/
|
||||
export class Pipeline {
|
||||
private static instance?: Pipeline
|
||||
|
||||
constructor() {
|
||||
if (Pipeline.instance) {
|
||||
return Pipeline.instance
|
||||
}
|
||||
Pipeline.instance = this
|
||||
}
|
||||
}
|
||||
|
||||
// Default singleton instance
|
||||
export const pipeline = new Pipeline()
|
||||
|
||||
// Backward compatibility aliases
|
||||
export const AugmentationPipeline = Pipeline
|
||||
export const augmentationPipeline = pipeline
|
||||
|
||||
// Factory functions
|
||||
export const createPipeline = async () => new Pipeline()
|
||||
export const createStreamingPipeline = async () => new Pipeline()
|
||||
|
||||
// Type aliases
|
||||
export type StreamlinedPipelineOptions = PipelineOptions
|
||||
export type PipelineResult<T> = { success: boolean; data: T; error?: string }
|
||||
export type StreamlinedPipelineResult<T> = PipelineResult<T>
|
||||
|
||||
// Execution mode alias
|
||||
export enum StreamlinedExecutionMode {
|
||||
SEQUENTIAL = 'sequential',
|
||||
PARALLEL = 'parallel',
|
||||
FIRST_SUCCESS = 'firstSuccess',
|
||||
FIRST_RESULT = 'firstResult',
|
||||
THREADED = 'threaded'
|
||||
}
|
||||
|
|
@ -287,207 +287,3 @@ export class EnhancedFileSystemClear {
|
|||
return result
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* Enhanced S3 bulk delete operations
|
||||
*/
|
||||
export class EnhancedS3Clear {
|
||||
constructor(
|
||||
private s3Client: any,
|
||||
private bucketName: string
|
||||
) {}
|
||||
|
||||
/**
|
||||
* Optimized bulk delete for S3 storage
|
||||
* Uses batch delete operations for maximum efficiency
|
||||
*/
|
||||
async clear(options: ClearOptions = {}): Promise<ClearResult> {
|
||||
const startTime = Date.now()
|
||||
const result: ClearResult = {
|
||||
success: false,
|
||||
itemsDeleted: { nouns: 0, verbs: 0, metadata: 0, system: 0 },
|
||||
duration: 0,
|
||||
errors: []
|
||||
}
|
||||
|
||||
try {
|
||||
// Safety checks
|
||||
if (options.confirmInstanceName) {
|
||||
// Extract instance name from bucket structure or prefix
|
||||
const bucketInfo = await this.getBucketInfo()
|
||||
if (bucketInfo.instanceName !== options.confirmInstanceName) {
|
||||
throw new Error(
|
||||
`Instance name mismatch: expected '${options.confirmInstanceName}', got '${bucketInfo.instanceName}'`
|
||||
)
|
||||
}
|
||||
}
|
||||
|
||||
// Dry run - just count objects
|
||||
if (options.dryRun) {
|
||||
return await this.performDryRun(options)
|
||||
}
|
||||
|
||||
// AWS S3 batch delete supports up to 1000 objects per request
|
||||
const batchSize = Math.min(options.batchSize || 1000, 1000)
|
||||
|
||||
// Delete with optimized batching
|
||||
const prefixes = [
|
||||
{ prefix: 'nouns/', key: 'nouns' as keyof typeof result.itemsDeleted },
|
||||
{ prefix: 'verbs/', key: 'verbs' as keyof typeof result.itemsDeleted },
|
||||
{ prefix: 'metadata/', key: 'metadata' as keyof typeof result.itemsDeleted },
|
||||
{ prefix: 'noun-metadata/', key: 'metadata' as keyof typeof result.itemsDeleted },
|
||||
{ prefix: 'verb-metadata/', key: 'metadata' as keyof typeof result.itemsDeleted },
|
||||
{ prefix: 'system/', key: 'system' as keyof typeof result.itemsDeleted },
|
||||
{ prefix: 'index/', key: 'system' as keyof typeof result.itemsDeleted }
|
||||
]
|
||||
|
||||
for (const { prefix, key } of prefixes) {
|
||||
const deleted = await this.clearPrefixOptimized(
|
||||
prefix,
|
||||
batchSize,
|
||||
(progress) => options.onProgress?.({
|
||||
...progress,
|
||||
stage: key === 'nouns' ? 'nouns' :
|
||||
key === 'verbs' ? 'verbs' :
|
||||
key === 'metadata' ? 'metadata' : 'system'
|
||||
})
|
||||
)
|
||||
result.itemsDeleted[key] += deleted
|
||||
}
|
||||
|
||||
result.success = true
|
||||
result.duration = Date.now() - startTime
|
||||
|
||||
} catch (error) {
|
||||
result.errors.push(error as Error)
|
||||
result.duration = Date.now() - startTime
|
||||
}
|
||||
|
||||
return result
|
||||
}
|
||||
|
||||
/**
|
||||
* High-performance prefix clearing using S3 batch delete
|
||||
*/
|
||||
private async clearPrefixOptimized(
|
||||
prefix: string,
|
||||
batchSize: number,
|
||||
onProgress?: (progress: Omit<ClearProgress, 'stage'>) => void
|
||||
): Promise<number> {
|
||||
const { ListObjectsV2Command, DeleteObjectsCommand } = await import('@aws-sdk/client-s3')
|
||||
|
||||
let totalDeleted = 0
|
||||
let continuationToken: string | undefined
|
||||
|
||||
do {
|
||||
// List objects with the prefix
|
||||
const listResponse = await this.s3Client.send(
|
||||
new ListObjectsV2Command({
|
||||
Bucket: this.bucketName,
|
||||
Prefix: prefix,
|
||||
MaxKeys: batchSize,
|
||||
ContinuationToken: continuationToken
|
||||
})
|
||||
)
|
||||
|
||||
if (!listResponse.Contents || listResponse.Contents.length === 0) {
|
||||
break
|
||||
}
|
||||
|
||||
// Prepare batch delete request
|
||||
const objectsToDelete = listResponse.Contents
|
||||
.filter((obj: any) => obj.Key)
|
||||
.map((obj: any) => ({ Key: obj.Key! }))
|
||||
|
||||
if (objectsToDelete.length > 0) {
|
||||
// Perform batch delete
|
||||
const deleteResponse = await this.s3Client.send(
|
||||
new DeleteObjectsCommand({
|
||||
Bucket: this.bucketName,
|
||||
Delete: {
|
||||
Objects: objectsToDelete,
|
||||
Quiet: false // Get detailed response
|
||||
}
|
||||
})
|
||||
)
|
||||
|
||||
const deletedCount = deleteResponse.Deleted?.length || 0
|
||||
totalDeleted += deletedCount
|
||||
|
||||
// Report any errors
|
||||
if (deleteResponse.Errors && deleteResponse.Errors.length > 0) {
|
||||
for (const error of deleteResponse.Errors) {
|
||||
console.warn(`Failed to delete ${error.Key}: ${error.Message}`)
|
||||
}
|
||||
}
|
||||
|
||||
// Report progress
|
||||
onProgress?.({
|
||||
totalItems: totalDeleted + (listResponse.IsTruncated ? 1000 : 0), // Estimate
|
||||
processedItems: totalDeleted,
|
||||
errors: deleteResponse.Errors?.length || 0
|
||||
})
|
||||
}
|
||||
|
||||
continuationToken = listResponse.NextContinuationToken
|
||||
|
||||
// Small delay to respect AWS rate limits
|
||||
await new Promise(resolve => setTimeout(resolve, 10))
|
||||
|
||||
} while (continuationToken)
|
||||
|
||||
return totalDeleted
|
||||
}
|
||||
|
||||
private async getBucketInfo(): Promise<{ instanceName: string }> {
|
||||
// Each Brainy instance has its own bucket with the same name as the instance
|
||||
// The bucket name IS the instance name
|
||||
return { instanceName: this.bucketName }
|
||||
}
|
||||
|
||||
private async performDryRun(options: ClearOptions): Promise<ClearResult> {
|
||||
const startTime = Date.now()
|
||||
const { ListObjectsV2Command } = await import('@aws-sdk/client-s3')
|
||||
|
||||
const result: ClearResult = {
|
||||
success: true,
|
||||
itemsDeleted: { nouns: 0, verbs: 0, metadata: 0, system: 0 },
|
||||
duration: 0,
|
||||
errors: []
|
||||
}
|
||||
|
||||
const countObjects = async (prefix: string): Promise<number> => {
|
||||
let count = 0
|
||||
let continuationToken: string | undefined
|
||||
|
||||
do {
|
||||
const response = await this.s3Client.send(
|
||||
new ListObjectsV2Command({
|
||||
Bucket: this.bucketName,
|
||||
Prefix: prefix,
|
||||
MaxKeys: 1000,
|
||||
ContinuationToken: continuationToken
|
||||
})
|
||||
)
|
||||
|
||||
count += response.KeyCount || 0
|
||||
continuationToken = response.NextContinuationToken
|
||||
} while (continuationToken)
|
||||
|
||||
return count
|
||||
}
|
||||
|
||||
result.itemsDeleted.nouns = await countObjects('nouns/')
|
||||
result.itemsDeleted.verbs = await countObjects('verbs/')
|
||||
result.itemsDeleted.metadata =
|
||||
await countObjects('metadata/') +
|
||||
await countObjects('noun-metadata/') +
|
||||
await countObjects('verb-metadata/')
|
||||
result.itemsDeleted.system =
|
||||
await countObjects('system/') +
|
||||
await countObjects('index/')
|
||||
|
||||
result.duration = Date.now() - startTime
|
||||
return result
|
||||
}
|
||||
}
|
||||
|
|
@ -18,7 +18,6 @@
|
|||
|
||||
import type { StorageAdapter } from '../coreTypes.js'
|
||||
import { MemoryStorage } from './adapters/memoryStorage.js'
|
||||
import { isBrowser } from '../utils/environment.js'
|
||||
import type { OperationConfig } from '../utils/operationUtils.js'
|
||||
|
||||
/**
|
||||
|
|
@ -80,11 +79,10 @@ function configureCOW(storage: any, options?: StorageOptions): void {
|
|||
/**
|
||||
* Resolve `StorageOptions` to a concrete storage adapter.
|
||||
*
|
||||
* - `'auto'` (default) picks `FileSystemStorage` when the runtime exposes
|
||||
* the Node `fs` API, otherwise falls back to `MemoryStorage`.
|
||||
* - `'auto'` (default) picks `FileSystemStorage`, falling back to
|
||||
* `MemoryStorage` only if filesystem init fails (e.g. no write permission).
|
||||
* - `forceMemoryStorage: true` returns `MemoryStorage` regardless.
|
||||
* - `forceFileSystemStorage: true` returns `FileSystemStorage` and throws in
|
||||
* the browser.
|
||||
* - `forceFileSystemStorage: true` returns `FileSystemStorage`.
|
||||
*/
|
||||
export async function createStorage(options: StorageOptions = {}): Promise<StorageAdapter> {
|
||||
const storage = await pickAdapter(options)
|
||||
|
|
@ -104,29 +102,17 @@ async function pickAdapter(options: StorageOptions): Promise<StorageAdapter> {
|
|||
}
|
||||
|
||||
if (requestedType === 'filesystem' || options.forceFileSystemStorage) {
|
||||
if (isBrowser()) {
|
||||
throw new Error(
|
||||
"Brainy: 'filesystem' storage is not available in browser environments. " +
|
||||
"Use `type: 'memory'` for in-browser brains, or run on a Node-like runtime."
|
||||
)
|
||||
}
|
||||
return await createFilesystemStorage(options)
|
||||
}
|
||||
|
||||
// 'auto': filesystem if we have Node-like fs, otherwise memory.
|
||||
if (!isBrowser()) {
|
||||
// 'auto': prefer filesystem, fall back to memory if init fails.
|
||||
try {
|
||||
return await createFilesystemStorage(options)
|
||||
} catch (err) {
|
||||
// Fall back to memory if filesystem init fails for any reason
|
||||
// (e.g. no write permission on rootDirectory).
|
||||
} catch {
|
||||
return new MemoryStorage()
|
||||
}
|
||||
}
|
||||
|
||||
return new MemoryStorage()
|
||||
}
|
||||
|
||||
async function createFilesystemStorage(options: StorageOptions): Promise<StorageAdapter> {
|
||||
const rootDir =
|
||||
options.rootDirectory ??
|
||||
|
|
|
|||
|
|
@ -1,74 +1,30 @@
|
|||
/**
|
||||
* Unified entry point for Brainy
|
||||
* This file exports everything from index.ts
|
||||
* Environment detection is handled here and made available to all components
|
||||
* @module unified
|
||||
* @description Unified entry point for Brainy. Re-exports the public surface
|
||||
* from `index.js` and surfaces a tiny runtime-environment object the rest of
|
||||
* the library can read.
|
||||
*
|
||||
* 8.0 supports Node-like runtimes only (Node.js, Bun, Deno); the browser
|
||||
* detection helpers and Worker Thread plumbing are gone.
|
||||
*/
|
||||
|
||||
// CRITICAL: The TensorFlow.js environment patch is now centralized in setup.ts
|
||||
// We import setup.ts below which applies the necessary patches
|
||||
|
||||
// CRITICAL: Import setup.js first to ensure TensorFlow.js environment patching
|
||||
// This MUST be the first import to prevent race conditions with TensorFlow.js initialization
|
||||
// Moving or removing this import will cause errors like "TextEncoder is not a constructor"
|
||||
// when the package is used in Node.js environments
|
||||
//
|
||||
// The setup.js file applies a patch that ensures TextEncoder/TextDecoder are properly
|
||||
// available to TensorFlow.js before it initializes its platform detection
|
||||
import './setup.js'
|
||||
|
||||
// Import environment detection functions
|
||||
import {
|
||||
isBrowser,
|
||||
isNode,
|
||||
isWebWorker,
|
||||
isThreadingAvailable,
|
||||
isThreadingAvailableAsync,
|
||||
areWorkerThreadsAvailable
|
||||
} from './utils/environment.js'
|
||||
import { isNode } from './utils/environment.js'
|
||||
|
||||
// Export environment information with lazy evaluation
|
||||
export const environment = {
|
||||
get isBrowser() {
|
||||
return isBrowser()
|
||||
},
|
||||
get isNode() {
|
||||
return isNode()
|
||||
},
|
||||
get isServerless() {
|
||||
return !isBrowser() && !isNode()
|
||||
},
|
||||
isWebWorker: function() {
|
||||
return isWebWorker()
|
||||
},
|
||||
get isThreadingAvailable() {
|
||||
return isThreadingAvailable()
|
||||
},
|
||||
isThreadingAvailableAsync: function() {
|
||||
return isThreadingAvailableAsync()
|
||||
},
|
||||
areWorkerThreadsAvailable: function() {
|
||||
return areWorkerThreadsAvailable()
|
||||
}
|
||||
}
|
||||
|
||||
// Make environment information available globally
|
||||
if (typeof globalThis !== 'undefined') {
|
||||
(globalThis as any).__ENV__ = environment
|
||||
;(globalThis as any).__ENV__ = environment
|
||||
}
|
||||
|
||||
// Log the detected environment
|
||||
console.log(
|
||||
`Brainy running in ${
|
||||
environment.isBrowser
|
||||
? 'browser'
|
||||
: environment.isNode
|
||||
? 'Node.js'
|
||||
: 'serverless/unknown'
|
||||
} environment`
|
||||
`Brainy running on ${environment.isNode ? 'Node-like runtime' : 'unknown runtime'}`
|
||||
)
|
||||
|
||||
// Re-export everything from index.ts
|
||||
export * from './index.js'
|
||||
|
||||
// Export the TensorFlow patch function for testing and manual use
|
||||
export { applyTensorFlowPatch } from './utils/textEncoding.js'
|
||||
|
|
|
|||
|
|
@ -1,474 +0,0 @@
|
|||
/**
|
||||
* Adaptive Socket Manager
|
||||
* Automatically manages socket pools and connection settings based on load patterns
|
||||
* Zero-configuration approach that learns and adapts to workload characteristics
|
||||
*/
|
||||
|
||||
import { Agent as HttpsAgent } from 'node:https'
|
||||
import { NodeHttpHandler } from '@smithy/node-http-handler'
|
||||
import { createModuleLogger } from './logger.js'
|
||||
|
||||
interface LoadMetrics {
|
||||
requestsPerSecond: number
|
||||
pendingRequests: number
|
||||
socketUtilization: number
|
||||
errorRate: number
|
||||
latencyP50: number
|
||||
latencyP95: number
|
||||
memoryUsage: number
|
||||
}
|
||||
|
||||
interface AdaptiveConfig {
|
||||
maxSockets: number
|
||||
maxFreeSockets: number
|
||||
keepAliveTimeout: number
|
||||
connectionTimeout: number
|
||||
socketTimeout: number
|
||||
batchSize: number
|
||||
}
|
||||
|
||||
/**
|
||||
* Adaptive Socket Manager that automatically scales based on load patterns
|
||||
*/
|
||||
export class AdaptiveSocketManager {
|
||||
private logger = createModuleLogger('AdaptiveSocketManager')
|
||||
|
||||
// Current configuration
|
||||
private config: AdaptiveConfig = {
|
||||
maxSockets: 100, // Start conservative
|
||||
maxFreeSockets: 20,
|
||||
keepAliveTimeout: 60000,
|
||||
connectionTimeout: 10000,
|
||||
socketTimeout: 60000,
|
||||
batchSize: 10
|
||||
}
|
||||
|
||||
// Performance tracking
|
||||
private metrics: LoadMetrics = {
|
||||
requestsPerSecond: 0,
|
||||
pendingRequests: 0,
|
||||
socketUtilization: 0,
|
||||
errorRate: 0,
|
||||
latencyP50: 0,
|
||||
latencyP95: 0,
|
||||
memoryUsage: 0
|
||||
}
|
||||
|
||||
// Historical data for learning
|
||||
private history: LoadMetrics[] = []
|
||||
private maxHistorySize = 100
|
||||
|
||||
// Adaptation state
|
||||
private lastAdaptationTime = 0
|
||||
private adaptationInterval = 5000 // Check every 5 seconds
|
||||
private consecutiveHighLoad = 0
|
||||
private consecutiveLowLoad = 0
|
||||
|
||||
// Request tracking
|
||||
private requestStartTimes = new Map<string, number>()
|
||||
private requestLatencies: number[] = []
|
||||
private errorCount = 0
|
||||
private successCount = 0
|
||||
private lastMetricReset = Date.now()
|
||||
|
||||
// Socket pool instances
|
||||
private currentAgent: HttpsAgent | null = null
|
||||
private currentHandler: NodeHttpHandler | null = null
|
||||
|
||||
/**
|
||||
* Get or create an optimized HTTP handler
|
||||
*/
|
||||
public getHttpHandler(): NodeHttpHandler {
|
||||
// Adapt configuration if needed
|
||||
this.adaptIfNeeded()
|
||||
|
||||
// Create new handler if configuration changed
|
||||
if (!this.currentHandler || this.shouldRecreateHandler()) {
|
||||
this.currentAgent = new HttpsAgent({
|
||||
keepAlive: true,
|
||||
maxSockets: this.config.maxSockets,
|
||||
maxFreeSockets: this.config.maxFreeSockets,
|
||||
timeout: this.config.keepAliveTimeout,
|
||||
scheduling: 'fifo' // Fair scheduling for high-volume scenarios
|
||||
})
|
||||
|
||||
this.currentHandler = new NodeHttpHandler({
|
||||
httpsAgent: this.currentAgent,
|
||||
connectionTimeout: this.config.connectionTimeout,
|
||||
socketTimeout: this.config.socketTimeout
|
||||
})
|
||||
|
||||
this.logger.debug('Created new HTTP handler with config:', this.config)
|
||||
}
|
||||
|
||||
return this.currentHandler
|
||||
}
|
||||
|
||||
/**
|
||||
* Get current batch size recommendation
|
||||
*/
|
||||
public getBatchSize(): number {
|
||||
this.adaptIfNeeded()
|
||||
return this.config.batchSize
|
||||
}
|
||||
|
||||
/**
|
||||
* Track request start
|
||||
*/
|
||||
public trackRequestStart(requestId: string): void {
|
||||
this.requestStartTimes.set(requestId, Date.now())
|
||||
this.metrics.pendingRequests++
|
||||
}
|
||||
|
||||
/**
|
||||
* Track request completion
|
||||
*/
|
||||
public trackRequestComplete(requestId: string, success: boolean): void {
|
||||
const startTime = this.requestStartTimes.get(requestId)
|
||||
if (startTime) {
|
||||
const latency = Date.now() - startTime
|
||||
this.requestLatencies.push(latency)
|
||||
this.requestStartTimes.delete(requestId)
|
||||
|
||||
// Keep latency array bounded
|
||||
if (this.requestLatencies.length > 1000) {
|
||||
this.requestLatencies = this.requestLatencies.slice(-500)
|
||||
}
|
||||
}
|
||||
|
||||
if (success) {
|
||||
this.successCount++
|
||||
} else {
|
||||
this.errorCount++
|
||||
}
|
||||
|
||||
this.metrics.pendingRequests = Math.max(0, this.metrics.pendingRequests - 1)
|
||||
}
|
||||
|
||||
/**
|
||||
* Check if we should adapt configuration
|
||||
*/
|
||||
private adaptIfNeeded(): void {
|
||||
const now = Date.now()
|
||||
if (now - this.lastAdaptationTime < this.adaptationInterval) {
|
||||
return
|
||||
}
|
||||
|
||||
this.lastAdaptationTime = now
|
||||
this.updateMetrics()
|
||||
this.analyzeAndAdapt()
|
||||
}
|
||||
|
||||
/**
|
||||
* Update current metrics
|
||||
*/
|
||||
private updateMetrics(): void {
|
||||
const now = Date.now()
|
||||
const timeSinceReset = (now - this.lastMetricReset) / 1000
|
||||
|
||||
// Calculate requests per second
|
||||
const totalRequests = this.successCount + this.errorCount
|
||||
this.metrics.requestsPerSecond = timeSinceReset > 0
|
||||
? totalRequests / timeSinceReset
|
||||
: 0
|
||||
|
||||
// Calculate error rate
|
||||
this.metrics.errorRate = totalRequests > 0
|
||||
? this.errorCount / totalRequests
|
||||
: 0
|
||||
|
||||
// Calculate latency percentiles
|
||||
if (this.requestLatencies.length > 0) {
|
||||
const sorted = [...this.requestLatencies].sort((a, b) => a - b)
|
||||
const p50Index = Math.floor(sorted.length * 0.5)
|
||||
const p95Index = Math.floor(sorted.length * 0.95)
|
||||
this.metrics.latencyP50 = sorted[p50Index] || 0
|
||||
this.metrics.latencyP95 = sorted[p95Index] || 0
|
||||
}
|
||||
|
||||
// Calculate socket utilization
|
||||
this.metrics.socketUtilization = this.metrics.pendingRequests / this.config.maxSockets
|
||||
|
||||
// Memory usage
|
||||
if (typeof process !== 'undefined' && process.memoryUsage) {
|
||||
const memUsage = process.memoryUsage()
|
||||
this.metrics.memoryUsage = memUsage.heapUsed / memUsage.heapTotal
|
||||
}
|
||||
|
||||
// Add to history
|
||||
this.history.push({ ...this.metrics })
|
||||
if (this.history.length > this.maxHistorySize) {
|
||||
this.history.shift()
|
||||
}
|
||||
|
||||
// Reset counters periodically
|
||||
if (timeSinceReset > 60) {
|
||||
this.lastMetricReset = now
|
||||
this.successCount = 0
|
||||
this.errorCount = 0
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* Analyze metrics and adapt configuration
|
||||
*/
|
||||
private analyzeAndAdapt(): void {
|
||||
const wasConfig = { ...this.config }
|
||||
|
||||
// Detect high load conditions
|
||||
const isHighLoad = this.detectHighLoad()
|
||||
const isLowLoad = this.detectLowLoad()
|
||||
const hasErrors = this.metrics.errorRate > 0.01 // More than 1% errors
|
||||
|
||||
if (isHighLoad) {
|
||||
this.consecutiveHighLoad++
|
||||
this.consecutiveLowLoad = 0
|
||||
|
||||
if (this.consecutiveHighLoad >= 2) { // Wait for 2 consecutive high load readings
|
||||
this.scaleUp()
|
||||
}
|
||||
} else if (isLowLoad) {
|
||||
this.consecutiveLowLoad++
|
||||
this.consecutiveHighLoad = 0
|
||||
|
||||
if (this.consecutiveLowLoad >= 6) { // Wait longer before scaling down
|
||||
this.scaleDown()
|
||||
}
|
||||
} else {
|
||||
// Reset counters if load is normal
|
||||
this.consecutiveHighLoad = Math.max(0, this.consecutiveHighLoad - 1)
|
||||
this.consecutiveLowLoad = Math.max(0, this.consecutiveLowLoad - 1)
|
||||
}
|
||||
|
||||
// Handle error conditions
|
||||
if (hasErrors) {
|
||||
this.handleErrors()
|
||||
}
|
||||
|
||||
// Log significant changes
|
||||
if (JSON.stringify(wasConfig) !== JSON.stringify(this.config)) {
|
||||
this.logger.info('Adapted configuration', {
|
||||
from: wasConfig,
|
||||
to: this.config,
|
||||
metrics: this.metrics
|
||||
})
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* Detect high load conditions
|
||||
*/
|
||||
private detectHighLoad(): boolean {
|
||||
return (
|
||||
this.metrics.socketUtilization > 0.7 || // Sockets heavily used
|
||||
this.metrics.pendingRequests > this.config.maxSockets * 0.8 || // Many pending requests
|
||||
this.metrics.latencyP95 > 5000 || // High latency
|
||||
this.metrics.requestsPerSecond > 100 // High request rate
|
||||
)
|
||||
}
|
||||
|
||||
/**
|
||||
* Detect low load conditions
|
||||
*/
|
||||
private detectLowLoad(): boolean {
|
||||
return (
|
||||
this.metrics.socketUtilization < 0.2 && // Sockets barely used
|
||||
this.metrics.pendingRequests < 5 && // Few pending requests
|
||||
this.metrics.latencyP95 < 1000 && // Low latency
|
||||
this.metrics.requestsPerSecond < 10 && // Low request rate
|
||||
this.metrics.memoryUsage < 0.5 // Low memory usage
|
||||
)
|
||||
}
|
||||
|
||||
/**
|
||||
* Scale up resources for high load
|
||||
*/
|
||||
private scaleUp(): void {
|
||||
// Increase socket limits progressively
|
||||
const scaleFactor = this.metrics.errorRate > 0.05 ? 1.5 : 2.0 // Scale more aggressively if no errors
|
||||
|
||||
this.config.maxSockets = Math.min(
|
||||
2000, // Hard limit to prevent resource exhaustion
|
||||
Math.ceil(this.config.maxSockets * scaleFactor)
|
||||
)
|
||||
|
||||
this.config.maxFreeSockets = Math.min(
|
||||
200,
|
||||
Math.ceil(this.config.maxSockets * 0.1) // Keep 10% as free sockets
|
||||
)
|
||||
|
||||
// Increase batch size for better throughput
|
||||
this.config.batchSize = Math.min(
|
||||
100,
|
||||
Math.ceil(this.config.batchSize * 1.5)
|
||||
)
|
||||
|
||||
// Adjust timeouts for high load
|
||||
this.config.keepAliveTimeout = 120000 // Keep connections alive longer
|
||||
this.config.connectionTimeout = 15000 // Allow more time for connections
|
||||
this.config.socketTimeout = 90000 // Allow more time for responses
|
||||
|
||||
this.logger.debug('Scaled up for high load', {
|
||||
sockets: this.config.maxSockets,
|
||||
batchSize: this.config.batchSize
|
||||
})
|
||||
}
|
||||
|
||||
/**
|
||||
* Scale down resources for low load
|
||||
*/
|
||||
private scaleDown(): void {
|
||||
// Only scale down if memory pressure is low
|
||||
if (this.metrics.memoryUsage > 0.7) {
|
||||
return
|
||||
}
|
||||
|
||||
// Decrease socket limits conservatively
|
||||
this.config.maxSockets = Math.max(
|
||||
50, // Minimum sockets
|
||||
Math.floor(this.config.maxSockets * 0.7)
|
||||
)
|
||||
|
||||
this.config.maxFreeSockets = Math.max(
|
||||
10,
|
||||
Math.floor(this.config.maxSockets * 0.2) // Keep 20% as free sockets
|
||||
)
|
||||
|
||||
// Decrease batch size
|
||||
this.config.batchSize = Math.max(
|
||||
5,
|
||||
Math.floor(this.config.batchSize * 0.7)
|
||||
)
|
||||
|
||||
// Adjust timeouts for low load
|
||||
this.config.keepAliveTimeout = 60000
|
||||
this.config.connectionTimeout = 10000
|
||||
this.config.socketTimeout = 60000
|
||||
|
||||
this.logger.debug('Scaled down for low load', {
|
||||
sockets: this.config.maxSockets,
|
||||
batchSize: this.config.batchSize
|
||||
})
|
||||
}
|
||||
|
||||
/**
|
||||
* Handle error conditions by adjusting configuration
|
||||
*/
|
||||
private handleErrors(): void {
|
||||
const errorRate = this.metrics.errorRate
|
||||
|
||||
if (errorRate > 0.1) { // More than 10% errors
|
||||
// Severe errors - back off aggressively
|
||||
this.config.maxSockets = Math.max(50, Math.floor(this.config.maxSockets * 0.5))
|
||||
this.config.batchSize = Math.max(1, Math.floor(this.config.batchSize * 0.3))
|
||||
this.config.connectionTimeout = Math.min(30000, this.config.connectionTimeout * 2)
|
||||
this.config.socketTimeout = Math.min(120000, this.config.socketTimeout * 2)
|
||||
|
||||
this.logger.warn('High error rate detected, backing off', {
|
||||
errorRate,
|
||||
newConfig: this.config
|
||||
})
|
||||
} else if (errorRate > 0.05) { // More than 5% errors
|
||||
// Moderate errors - reduce load slightly
|
||||
this.config.batchSize = Math.max(1, Math.floor(this.config.batchSize * 0.7))
|
||||
this.config.connectionTimeout = Math.min(20000, this.config.connectionTimeout * 1.2)
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* Check if we should recreate the handler
|
||||
*/
|
||||
private shouldRecreateHandler(): boolean {
|
||||
if (!this.currentAgent) return true
|
||||
|
||||
// Recreate if socket configuration changed significantly
|
||||
const currentMaxSockets = (this.currentAgent as any).maxSockets
|
||||
const socketsDiff = Math.abs(currentMaxSockets - this.config.maxSockets)
|
||||
|
||||
return socketsDiff > currentMaxSockets * 0.5 // 50% change threshold
|
||||
}
|
||||
|
||||
/**
|
||||
* Get current configuration (for monitoring)
|
||||
*/
|
||||
public getConfig(): Readonly<AdaptiveConfig> {
|
||||
return { ...this.config }
|
||||
}
|
||||
|
||||
/**
|
||||
* Get current metrics (for monitoring)
|
||||
*/
|
||||
public getMetrics(): Readonly<LoadMetrics> {
|
||||
return { ...this.metrics }
|
||||
}
|
||||
|
||||
/**
|
||||
* Predict optimal configuration based on historical data
|
||||
*/
|
||||
public predictOptimalConfig(): AdaptiveConfig {
|
||||
if (this.history.length < 10) {
|
||||
return this.config // Not enough data to predict
|
||||
}
|
||||
|
||||
// Analyze recent history
|
||||
const recentHistory = this.history.slice(-20)
|
||||
const avgRPS = recentHistory.reduce((sum, m) => sum + m.requestsPerSecond, 0) / recentHistory.length
|
||||
const maxRPS = Math.max(...recentHistory.map(m => m.requestsPerSecond))
|
||||
const avgLatency = recentHistory.reduce((sum, m) => sum + m.latencyP95, 0) / recentHistory.length
|
||||
|
||||
// Predict optimal socket count based on request patterns
|
||||
const optimalSockets = Math.min(2000, Math.max(50, Math.ceil(maxRPS * 2)))
|
||||
|
||||
// Predict optimal batch size based on latency
|
||||
const optimalBatchSize = avgLatency < 1000 ? 50 : avgLatency < 3000 ? 20 : 10
|
||||
|
||||
return {
|
||||
maxSockets: optimalSockets,
|
||||
maxFreeSockets: Math.ceil(optimalSockets * 0.15),
|
||||
keepAliveTimeout: avgRPS > 50 ? 120000 : 60000,
|
||||
connectionTimeout: avgLatency > 3000 ? 20000 : 10000,
|
||||
socketTimeout: avgLatency > 3000 ? 90000 : 60000,
|
||||
batchSize: optimalBatchSize
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* Reset to default configuration
|
||||
*/
|
||||
public reset(): void {
|
||||
this.config = {
|
||||
maxSockets: 100,
|
||||
maxFreeSockets: 20,
|
||||
keepAliveTimeout: 60000,
|
||||
connectionTimeout: 10000,
|
||||
socketTimeout: 60000,
|
||||
batchSize: 10
|
||||
}
|
||||
|
||||
this.consecutiveHighLoad = 0
|
||||
this.consecutiveLowLoad = 0
|
||||
this.history = []
|
||||
this.requestLatencies = []
|
||||
this.errorCount = 0
|
||||
this.successCount = 0
|
||||
|
||||
// Force recreation of handler
|
||||
this.currentAgent = null
|
||||
this.currentHandler = null
|
||||
|
||||
this.logger.info('Reset to default configuration')
|
||||
}
|
||||
}
|
||||
|
||||
// Global singleton instance
|
||||
let globalSocketManager: AdaptiveSocketManager | null = null
|
||||
|
||||
/**
|
||||
* Get the global socket manager instance
|
||||
*/
|
||||
export function getGlobalSocketManager(): AdaptiveSocketManager {
|
||||
if (!globalSocketManager) {
|
||||
globalSocketManager = new AdaptiveSocketManager()
|
||||
}
|
||||
return globalSocketManager
|
||||
}
|
||||
|
|
@ -3,16 +3,16 @@
|
|||
* Detects environment, resources, and data patterns to provide optimal settings
|
||||
*/
|
||||
|
||||
import { isBrowser, isNode, isThreadingAvailable } from './environment.js'
|
||||
import { isNode } from './environment.js'
|
||||
|
||||
export interface AutoConfigResult {
|
||||
// Environment details
|
||||
environment: 'browser' | 'nodejs' | 'serverless' | 'unknown'
|
||||
environment: 'nodejs' | 'serverless' | 'unknown'
|
||||
|
||||
// Resource detection
|
||||
availableMemory: number // bytes
|
||||
cpuCores: number
|
||||
threadingAvailable: boolean
|
||||
|
||||
|
||||
// Storage capabilities
|
||||
persistentStorageAvailable: boolean
|
||||
|
|
@ -201,11 +201,7 @@ export class AutoConfiguration {
|
|||
/**
|
||||
* Detect the current runtime environment
|
||||
*/
|
||||
private detectEnvironment(): 'browser' | 'nodejs' | 'serverless' | 'unknown' {
|
||||
if (isBrowser()) {
|
||||
return 'browser'
|
||||
}
|
||||
|
||||
private detectEnvironment(): 'nodejs' | 'serverless' | 'unknown' {
|
||||
if (isNode()) {
|
||||
// Check for serverless environment indicators
|
||||
if (process.env.AWS_LAMBDA_FUNCTION_NAME ||
|
||||
|
|
@ -226,40 +222,21 @@ export class AutoConfiguration {
|
|||
private async detectResources(): Promise<{
|
||||
availableMemory: number
|
||||
cpuCores: number
|
||||
threadingAvailable: boolean
|
||||
}> {
|
||||
let availableMemory = 2 * 1024 * 1024 * 1024 // Default 2GB
|
||||
let cpuCores = 4 // Default 4 cores
|
||||
|
||||
// Browser memory detection
|
||||
if (isBrowser()) {
|
||||
// @ts-ignore - navigator.deviceMemory is experimental
|
||||
if (navigator.deviceMemory) {
|
||||
// @ts-ignore
|
||||
availableMemory = navigator.deviceMemory * 1024 * 1024 * 1024 * 0.3 // Use 30% of device memory
|
||||
} else {
|
||||
availableMemory = 512 * 1024 * 1024 // Conservative 512MB for browsers
|
||||
}
|
||||
|
||||
cpuCores = navigator.hardwareConcurrency || 4
|
||||
}
|
||||
|
||||
// Node.js memory detection
|
||||
if (isNode()) {
|
||||
try {
|
||||
const os = await import('node:os')
|
||||
availableMemory = os.totalmem() * 0.7 // Use 70% of total memory
|
||||
cpuCores = os.cpus().length
|
||||
} catch (error) {
|
||||
} catch {
|
||||
// Fallback to defaults
|
||||
}
|
||||
}
|
||||
|
||||
return {
|
||||
availableMemory,
|
||||
cpuCores,
|
||||
threadingAvailable: isThreadingAvailable()
|
||||
}
|
||||
return { availableMemory, cpuCores }
|
||||
}
|
||||
|
||||
/**
|
||||
|
|
@ -272,17 +249,10 @@ export class AutoConfiguration {
|
|||
let persistentStorageAvailable = false
|
||||
let s3StorageDetected = s3Hint || false
|
||||
|
||||
if (isBrowser()) {
|
||||
// Check for OPFS support
|
||||
persistentStorageAvailable = 'navigator' in globalThis &&
|
||||
'storage' in navigator &&
|
||||
'getDirectory' in navigator.storage
|
||||
}
|
||||
|
||||
if (isNode()) {
|
||||
persistentStorageAvailable = true // Always available in Node.js
|
||||
|
||||
// Check for AWS SDK or S3 environment variables
|
||||
persistentStorageAvailable = true
|
||||
// 8.0 dropped cloud storage adapters; the s3StorageDetected flag is
|
||||
// preserved as a hint for operator backup tooling, not adapter wiring.
|
||||
s3StorageDetected = s3Hint ||
|
||||
!!(process.env.AWS_ACCESS_KEY_ID && process.env.AWS_SECRET_ACCESS_KEY) ||
|
||||
!!(process.env.S3_BUCKET_NAME)
|
||||
|
|
@ -311,7 +281,7 @@ export class AutoConfiguration {
|
|||
maxMemoryUsage: memoryBudget,
|
||||
targetSearchLatency: 150,
|
||||
enablePartitioning: datasetSize > 25000,
|
||||
enableCompression: environment === 'browser' || memoryBudget < 2 * 1024 * 1024 * 1024,
|
||||
enableCompression: memoryBudget < 2 * 1024 * 1024 * 1024,
|
||||
enableDistributedSearch: resources.cpuCores > 2 && datasetSize > 50000,
|
||||
enablePredictiveCaching: true,
|
||||
partitionStrategy: 'semantic' as const,
|
||||
|
|
@ -321,17 +291,6 @@ export class AutoConfiguration {
|
|||
|
||||
// Environment-specific adjustments
|
||||
switch (environment) {
|
||||
case 'browser':
|
||||
config = {
|
||||
...config,
|
||||
maxMemoryUsage: Math.min(memoryBudget, 1024 * 1024 * 1024), // Cap at 1GB
|
||||
targetSearchLatency: 200, // More lenient for browsers
|
||||
enableCompression: true, // Always enable for browsers
|
||||
maxNodesPerPartition: 25000, // Smaller partitions
|
||||
semanticClusters: 4 // Fewer clusters to save memory
|
||||
}
|
||||
break
|
||||
|
||||
case 'serverless':
|
||||
config = {
|
||||
...config,
|
||||
|
|
@ -437,7 +396,6 @@ export class AutoConfiguration {
|
|||
const environment = this.detectEnvironment()
|
||||
|
||||
switch (environment) {
|
||||
case 'browser': return 10000
|
||||
case 'serverless': return 50000
|
||||
case 'nodejs': return 100000
|
||||
default: return 25000
|
||||
|
|
|
|||
|
|
@ -5,8 +5,6 @@
|
|||
*/
|
||||
|
||||
import { DistanceFunction, Vector } from '../coreTypes.js'
|
||||
import { executeInThread } from './workerUtils.js'
|
||||
import { isThreadingAvailable } from './environment.js'
|
||||
|
||||
/**
|
||||
* Calculates the Euclidean distance between two vectors
|
||||
|
|
|
|||
|
|
@ -1,25 +1,17 @@
|
|||
/**
|
||||
* Utility functions for environment detection
|
||||
* @module utils/environment
|
||||
* @description Runtime environment detection helpers. 8.0 supports Node.js,
|
||||
* Bun, and Deno on the server side only — browsers were dropped from the
|
||||
* support matrix in 8.0, so the helpers here assume a Node-like runtime.
|
||||
*/
|
||||
|
||||
/**
|
||||
* Check if code is running in a browser environment
|
||||
* @deprecated Browser support is deprecated and will be removed in v8.0. Use Node.js or Bun instead.
|
||||
*/
|
||||
export function isBrowser(): boolean {
|
||||
return typeof window !== 'undefined' && typeof document !== 'undefined'
|
||||
}
|
||||
|
||||
/**
|
||||
* Check if code is running in a Node.js environment
|
||||
* @description True when running inside a Node-compatible runtime
|
||||
* (Node.js, Bun, Deno node-compat). 8.0 has no browser path, so this is
|
||||
* effectively always true in a deployed brain — the function is kept for
|
||||
* defensive checks at runtime boundaries.
|
||||
*/
|
||||
export function isNode(): boolean {
|
||||
// If browser environment is detected, prioritize it over Node.js
|
||||
// This handles cases like jsdom where both window and process exist
|
||||
if (isBrowser()) {
|
||||
return false
|
||||
}
|
||||
|
||||
return (
|
||||
typeof process !== 'undefined' &&
|
||||
process.versions != null &&
|
||||
|
|
@ -28,156 +20,59 @@ export function isNode(): boolean {
|
|||
}
|
||||
|
||||
/**
|
||||
* Check if code is running in a Web Worker environment
|
||||
*/
|
||||
export function isWebWorker(): boolean {
|
||||
return (
|
||||
typeof self === 'object' &&
|
||||
self.constructor &&
|
||||
self.constructor.name === 'DedicatedWorkerGlobalScope'
|
||||
)
|
||||
}
|
||||
|
||||
/**
|
||||
* Check if Web Workers are available in the current environment
|
||||
* @deprecated Browser support is deprecated and will be removed in v8.0. Use Node.js Worker Threads instead.
|
||||
*/
|
||||
export function areWebWorkersAvailable(): boolean {
|
||||
return isBrowser() && typeof Worker !== 'undefined'
|
||||
}
|
||||
|
||||
/**
|
||||
* Check if Worker Threads are available in the current environment (Node.js)
|
||||
*/
|
||||
export async function areWorkerThreadsAvailable(): Promise<boolean> {
|
||||
if (!isNode()) return false
|
||||
|
||||
try {
|
||||
// Use dynamic import to avoid errors in browser environments
|
||||
await import('worker_threads')
|
||||
return true
|
||||
} catch (e) {
|
||||
return false
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* Synchronous version that doesn't actually try to load the module
|
||||
* This is safer in ES module environments
|
||||
*/
|
||||
export function areWorkerThreadsAvailableSync(): boolean {
|
||||
if (!isNode()) return false
|
||||
|
||||
// In Node.js 22+, worker_threads is always available
|
||||
return parseInt(process.versions.node.split('.')[0]) >= 22
|
||||
}
|
||||
|
||||
/**
|
||||
* Determine if threading is available in the current environment
|
||||
* Returns true if either Web Workers (browser) or Worker Threads (Node.js) are available
|
||||
*/
|
||||
export function isThreadingAvailable(): boolean {
|
||||
return areWebWorkersAvailable() || areWorkerThreadsAvailableSync()
|
||||
}
|
||||
|
||||
/**
|
||||
* Async version of isThreadingAvailable
|
||||
*/
|
||||
export async function isThreadingAvailableAsync(): Promise<boolean> {
|
||||
return areWebWorkersAvailable() || (await areWorkerThreadsAvailable())
|
||||
}
|
||||
|
||||
/**
|
||||
* Auto-detect production environment to minimize logging costs
|
||||
* @description Auto-detect a production deployment. Returns true when any
|
||||
* common managed-runtime indicator is present (Cloud Run, Lambda, Azure
|
||||
* Functions, Vercel, Netlify, Heroku, Railway, Fly.io, Docker prod), or
|
||||
* when `NODE_ENV=production`. Used to silence verbose logging in prod.
|
||||
*/
|
||||
export function isProductionEnvironment(): boolean {
|
||||
// Node.js environment detection
|
||||
if (isNode()) {
|
||||
// Check common production environment indicators
|
||||
if (!isNode()) return false
|
||||
|
||||
const nodeEnv = process.env.NODE_ENV?.toLowerCase()
|
||||
if (nodeEnv === 'production' || nodeEnv === 'prod') return true
|
||||
|
||||
// Google Cloud Run detection
|
||||
if (process.env.K_SERVICE || process.env.GOOGLE_CLOUD_PROJECT) return true
|
||||
|
||||
// AWS Lambda detection
|
||||
if (process.env.AWS_LAMBDA_FUNCTION_NAME || process.env.AWS_EXECUTION_ENV) return true
|
||||
|
||||
// Azure Functions detection
|
||||
if (process.env.AZURE_FUNCTIONS_ENVIRONMENT || process.env.WEBSITE_SITE_NAME) return true
|
||||
|
||||
// Vercel detection
|
||||
if (process.env.VERCEL || process.env.VERCEL_ENV === 'production') return true
|
||||
|
||||
// Netlify detection
|
||||
if (process.env.NETLIFY && process.env.CONTEXT === 'production') return true
|
||||
|
||||
// Heroku detection
|
||||
if (process.env.DYNO && process.env.NODE_ENV !== 'development') return true
|
||||
|
||||
// Railway detection
|
||||
if (process.env.RAILWAY_ENVIRONMENT === 'production') return true
|
||||
|
||||
// Fly.io detection
|
||||
if (process.env.FLY_APP_NAME && process.env.FLY_REGION) return true
|
||||
|
||||
// Docker in production (common patterns)
|
||||
if (process.env.DOCKER_ENV === 'production' || process.env.ENVIRONMENT === 'production') return true
|
||||
|
||||
// Generic production indicators
|
||||
if (process.env.PROD === 'true' || process.env.PRODUCTION === 'true') return true
|
||||
}
|
||||
|
||||
// Browser environment - assume development unless explicitly production
|
||||
if (isBrowser()) {
|
||||
// Check for production domain patterns
|
||||
const hostname = window?.location?.hostname
|
||||
if (hostname) {
|
||||
// Avoid logging on production domains
|
||||
if (hostname.includes('.com') || hostname.includes('.org') || hostname.includes('.net')) {
|
||||
return !hostname.includes('localhost') && !hostname.includes('127.0.0.1') && !hostname.includes('dev')
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
return false
|
||||
}
|
||||
|
||||
/**
|
||||
* Get appropriate log level based on environment
|
||||
* @description Resolve a log level from BRAINY_LOG_LEVEL, NODE_ENV, and
|
||||
* deployment-environment heuristics. Production defaults to 'error' so a
|
||||
* busy app doesn't pay for verbose logging.
|
||||
*/
|
||||
export function getLogLevel(): 'silent' | 'error' | 'warn' | 'info' | 'verbose' {
|
||||
// Explicit log level override
|
||||
const explicitLevel = process.env.BRAINY_LOG_LEVEL?.toLowerCase()
|
||||
if (explicitLevel && ['silent', 'error', 'warn', 'info', 'verbose'].includes(explicitLevel)) {
|
||||
return explicitLevel as 'silent' | 'error' | 'warn' | 'info' | 'verbose'
|
||||
}
|
||||
|
||||
// Auto-detect based on environment
|
||||
if (isProductionEnvironment()) {
|
||||
return 'error' // Only log errors in production to minimize costs
|
||||
}
|
||||
if (isProductionEnvironment()) return 'error'
|
||||
|
||||
// Development environments get more verbose logging
|
||||
if (process.env.NODE_ENV === 'development' || process.env.NODE_ENV === 'dev') {
|
||||
return 'verbose'
|
||||
}
|
||||
|
||||
// Test environments should be quieter
|
||||
if (process.env.NODE_ENV === 'test') {
|
||||
return 'warn'
|
||||
}
|
||||
if (process.env.NODE_ENV === 'test') return 'warn'
|
||||
|
||||
// Default to info level
|
||||
return 'info'
|
||||
}
|
||||
|
||||
/**
|
||||
* Check if logging should be enabled for a given level
|
||||
* @description True when a log message at the requested level should be
|
||||
* emitted given the current configured log level. Returns false in 'silent'.
|
||||
*/
|
||||
export function shouldLog(level: 'error' | 'warn' | 'info' | 'verbose'): boolean {
|
||||
const currentLevel = getLogLevel()
|
||||
|
||||
if (currentLevel === 'silent') return false
|
||||
|
||||
const levels = ['error', 'warn', 'info', 'verbose']
|
||||
|
|
|
|||
|
|
@ -1,6 +1,5 @@
|
|||
export * from './distance.js'
|
||||
export * from './embedding.js'
|
||||
export * from './workerUtils.js'
|
||||
export * from './statisticsCollector.js'
|
||||
export * from './jsonProcessing.js'
|
||||
export * from './fieldNameTracking.js'
|
||||
|
|
|
|||
|
|
@ -5,7 +5,6 @@
|
|||
*/
|
||||
|
||||
import { createModuleLogger } from './logger.js'
|
||||
import { getGlobalSocketManager } from './adaptiveSocketManager.js'
|
||||
import { getGlobalBackpressure } from './adaptiveBackpressure.js'
|
||||
|
||||
interface PerformanceMetrics {
|
||||
|
|
@ -215,9 +214,10 @@ export class PerformanceMonitor {
|
|||
}
|
||||
}
|
||||
|
||||
// Get metrics from socket manager
|
||||
const socketMetrics = getGlobalSocketManager().getMetrics()
|
||||
this.metrics.socketUtilization = socketMetrics.socketUtilization
|
||||
// Socket-pool metrics were tied to the dropped cloud HTTP handler; not
|
||||
// applicable to filesystem-only 8.0 deployments.
|
||||
this.metrics.socketUtilization = 0
|
||||
|
||||
|
||||
// Get metrics from backpressure system
|
||||
const backpressureStatus = getGlobalBackpressure().getStatus()
|
||||
|
|
@ -431,14 +431,12 @@ export class PerformanceMonitor {
|
|||
metrics: PerformanceMetrics
|
||||
trends: PerformanceTrend[]
|
||||
recommendations: string[]
|
||||
socketConfig: any
|
||||
backpressureStatus: any
|
||||
} {
|
||||
return {
|
||||
metrics: this.getMetrics(),
|
||||
trends: this.getTrends(),
|
||||
recommendations: this.getRecommendations(),
|
||||
socketConfig: getGlobalSocketManager().getConfig(),
|
||||
backpressureStatus: getGlobalBackpressure().getStatus()
|
||||
}
|
||||
}
|
||||
|
|
|
|||
|
|
@ -13,15 +13,7 @@ let patchApplied = false
|
|||
* Simplified version for Transformers.js/ONNX Runtime
|
||||
*/
|
||||
export async function applyTensorFlowPatch(): Promise<void> {
|
||||
// Apply patches for all non-browser environments that might need TextEncoder/TextDecoder
|
||||
const isBrowserEnv = typeof window !== 'undefined' && typeof document !== 'undefined'
|
||||
if (isBrowserEnv || patchApplied) {
|
||||
return // Browser environments don't need these patches, and don't patch twice
|
||||
}
|
||||
|
||||
if (!isNode()) {
|
||||
return // Only patch Node.js environments
|
||||
}
|
||||
if (patchApplied || !isNode()) return
|
||||
|
||||
try {
|
||||
console.log('Brainy: Applying TextEncoder/TextDecoder patch for Node.js')
|
||||
|
|
|
|||
|
|
@ -1,512 +0,0 @@
|
|||
/**
|
||||
* Utility functions for executing functions in Worker Threads (Node.js) or Web Workers (Browser)
|
||||
* This implementation leverages Node.js 22's stable Worker Threads API for maximum reliability
|
||||
*/
|
||||
|
||||
import { isBrowser, isNode } from './environment.js'
|
||||
import { prodLog } from './logger.js'
|
||||
|
||||
// Worker pool to reuse workers
|
||||
const workerPool: Map<string, any> = new Map()
|
||||
const MAX_POOL_SIZE = 4 // Adjust based on system capabilities
|
||||
|
||||
/**
|
||||
* Execute a function in a separate thread
|
||||
*
|
||||
* @param fnString The function to execute as a string
|
||||
* @param args The arguments to pass to the function
|
||||
* @returns A promise that resolves with the result of the function
|
||||
*/
|
||||
export function executeInThread<T>(fnString: string, args: any): Promise<T> {
|
||||
if (isNode()) {
|
||||
return executeInNodeWorker<T>(fnString, args)
|
||||
} else if (isBrowser() && typeof window !== 'undefined' && window.Worker) {
|
||||
return executeInWebWorker<T>(fnString, args)
|
||||
} else {
|
||||
// Fallback to main thread execution
|
||||
try {
|
||||
// Try different approaches to create a function from string
|
||||
let fn
|
||||
try {
|
||||
// First try with 'return' prefix
|
||||
fn = new Function('return ' + fnString)()
|
||||
} catch (functionError) {
|
||||
console.warn(
|
||||
'Fallback: Error creating function with return syntax, trying alternative approaches',
|
||||
functionError
|
||||
)
|
||||
|
||||
try {
|
||||
// Try wrapping in parentheses for function expressions
|
||||
fn = new Function('return (' + fnString + ')')()
|
||||
} catch (wrapError) {
|
||||
console.warn(
|
||||
'Fallback: Error creating function with parentheses wrapping',
|
||||
wrapError
|
||||
)
|
||||
|
||||
try {
|
||||
// Try direct approach for named functions
|
||||
fn = new Function(fnString)()
|
||||
} catch (directError) {
|
||||
console.warn(
|
||||
'Fallback: Direct approach failed, trying with function wrapper',
|
||||
directError
|
||||
)
|
||||
|
||||
try {
|
||||
// Try wrapping in a function that returns the function expression
|
||||
fn = new Function(
|
||||
'return function(args) { return (' + fnString + ')(args); }'
|
||||
)()
|
||||
} catch (wrapperError) {
|
||||
console.error(
|
||||
'Fallback: All approaches to create function failed',
|
||||
wrapperError
|
||||
)
|
||||
throw new Error(
|
||||
'Failed to create function from string: ' +
|
||||
(functionError as Error).message
|
||||
)
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
return Promise.resolve(fn(args) as T)
|
||||
} catch (error) {
|
||||
return Promise.reject(error)
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* Execute a function in a Node.js Worker Thread
|
||||
* Optimized for Node.js 22 LTS with stable Worker Threads performance
|
||||
*/
|
||||
function executeInNodeWorker<T>(fnString: string, args: any): Promise<T> {
|
||||
return new Promise<T>((resolve, reject) => {
|
||||
try {
|
||||
// Dynamically import worker_threads (Node.js only)
|
||||
import('node:worker_threads')
|
||||
.then(({ Worker, isMainThread, parentPort, workerData }) => {
|
||||
if (!isMainThread && parentPort) {
|
||||
// We're inside a worker, execute the function
|
||||
const fn = new Function('return ' + workerData.fnString)()
|
||||
const result = fn(workerData.args)
|
||||
parentPort.postMessage({ result })
|
||||
return
|
||||
}
|
||||
|
||||
// Get a worker from the pool or create a new one
|
||||
const workerId = `worker-${Math.random().toString(36).substring(2, 9)}`
|
||||
let worker: any
|
||||
|
||||
if (workerPool.size < MAX_POOL_SIZE) {
|
||||
// Create a new worker
|
||||
worker = new Worker(
|
||||
`
|
||||
import { parentPort, workerData } from 'node:worker_threads';
|
||||
|
||||
// Add TensorFlow.js platform patch for Node.js
|
||||
if (typeof global !== 'undefined') {
|
||||
try {
|
||||
// Define a custom PlatformNode class
|
||||
class PlatformNode {
|
||||
constructor() {
|
||||
// Create a util object with necessary methods
|
||||
this.util = {
|
||||
// Add isFloat32Array and isTypedArray directly to util
|
||||
isFloat32Array: (arr) => {
|
||||
return !!(
|
||||
arr instanceof Float32Array ||
|
||||
(arr &&
|
||||
Object.prototype.toString.call(arr) === '[object Float32Array]')
|
||||
);
|
||||
},
|
||||
isTypedArray: (arr) => {
|
||||
return !!(ArrayBuffer.isView(arr) && !(arr instanceof DataView));
|
||||
},
|
||||
// Use native TextEncoder and TextDecoder
|
||||
TextEncoder: TextEncoder,
|
||||
TextDecoder: TextDecoder
|
||||
};
|
||||
|
||||
// Initialize encoders using native constructors
|
||||
this.textEncoder = new TextEncoder();
|
||||
this.textDecoder = new TextDecoder();
|
||||
}
|
||||
|
||||
// Define isFloat32Array directly on the instance
|
||||
isFloat32Array(arr) {
|
||||
return !!(
|
||||
arr instanceof Float32Array ||
|
||||
(arr && Object.prototype.toString.call(arr) === '[object Float32Array]')
|
||||
);
|
||||
}
|
||||
|
||||
// Define isTypedArray directly on the instance
|
||||
isTypedArray(arr) {
|
||||
return !!(ArrayBuffer.isView(arr) && !(arr instanceof DataView));
|
||||
}
|
||||
}
|
||||
|
||||
// Assign the PlatformNode class to the global object
|
||||
global.PlatformNode = PlatformNode;
|
||||
|
||||
// Also create an instance and assign it to global.platformNode
|
||||
global.platformNode = new PlatformNode();
|
||||
|
||||
// Ensure global.util exists and has the necessary methods
|
||||
if (!global.util) {
|
||||
global.util = {};
|
||||
}
|
||||
|
||||
// Add isFloat32Array method if it doesn't exist
|
||||
if (!global.util.isFloat32Array) {
|
||||
global.util.isFloat32Array = (arr) => {
|
||||
return !!(
|
||||
arr instanceof Float32Array ||
|
||||
(arr && Object.prototype.toString.call(arr) === '[object Float32Array]')
|
||||
);
|
||||
};
|
||||
}
|
||||
|
||||
// Add isTypedArray method if it doesn't exist
|
||||
if (!global.util.isTypedArray) {
|
||||
global.util.isTypedArray = (arr) => {
|
||||
return !!(ArrayBuffer.isView(arr) && !(arr instanceof DataView));
|
||||
};
|
||||
}
|
||||
} catch (error) {
|
||||
console.warn('Failed to apply TensorFlow.js platform patch:', error);
|
||||
}
|
||||
}
|
||||
|
||||
const fn = new Function('return ' + workerData.fnString)();
|
||||
const result = fn(workerData.args);
|
||||
parentPort.postMessage({ result });
|
||||
`,
|
||||
{
|
||||
eval: true,
|
||||
workerData: { fnString, args }
|
||||
}
|
||||
)
|
||||
|
||||
workerPool.set(workerId, worker)
|
||||
} else {
|
||||
// Reuse an existing worker
|
||||
const poolKeys = Array.from(workerPool.keys())
|
||||
const randomKey =
|
||||
poolKeys[Math.floor(Math.random() * poolKeys.length)]
|
||||
worker = workerPool.get(randomKey)
|
||||
|
||||
// Terminate and recreate if the worker is busy
|
||||
if (worker._busy) {
|
||||
worker.terminate()
|
||||
worker = new Worker(
|
||||
`
|
||||
import { parentPort, workerData } from 'node:worker_threads';
|
||||
|
||||
// Add TensorFlow.js platform patch for Node.js
|
||||
if (typeof global !== 'undefined') {
|
||||
try {
|
||||
// Define a custom PlatformNode class
|
||||
class PlatformNode {
|
||||
constructor() {
|
||||
// Create a util object with necessary methods
|
||||
this.util = {
|
||||
// Use native TextEncoder and TextDecoder
|
||||
TextEncoder: TextEncoder,
|
||||
TextDecoder: TextDecoder
|
||||
};
|
||||
|
||||
// Initialize encoders using native constructors
|
||||
this.textEncoder = new TextEncoder();
|
||||
this.textDecoder = new TextDecoder();
|
||||
}
|
||||
|
||||
// Define isFloat32Array directly on the instance
|
||||
isFloat32Array(arr) {
|
||||
return !!(
|
||||
arr instanceof Float32Array ||
|
||||
(arr && Object.prototype.toString.call(arr) === '[object Float32Array]')
|
||||
);
|
||||
}
|
||||
|
||||
// Define isTypedArray directly on the instance
|
||||
isTypedArray(arr) {
|
||||
return !!(ArrayBuffer.isView(arr) && !(arr instanceof DataView));
|
||||
}
|
||||
}
|
||||
|
||||
// Assign the PlatformNode class to the global object
|
||||
global.PlatformNode = PlatformNode;
|
||||
|
||||
// Also create an instance and assign it to global.platformNode
|
||||
global.platformNode = new PlatformNode();
|
||||
|
||||
// Ensure global.util exists and has the necessary methods
|
||||
if (!global.util) {
|
||||
global.util = {};
|
||||
}
|
||||
|
||||
// Add isFloat32Array method if it doesn't exist
|
||||
if (!global.util.isFloat32Array) {
|
||||
global.util.isFloat32Array = (arr) => {
|
||||
return !!(
|
||||
arr instanceof Float32Array ||
|
||||
(arr && Object.prototype.toString.call(arr) === '[object Float32Array]')
|
||||
);
|
||||
};
|
||||
}
|
||||
|
||||
// Add isTypedArray method if it doesn't exist
|
||||
if (!global.util.isTypedArray) {
|
||||
global.util.isTypedArray = (arr) => {
|
||||
return !!(ArrayBuffer.isView(arr) && !(arr instanceof DataView));
|
||||
};
|
||||
}
|
||||
} catch (error) {
|
||||
console.warn('Failed to apply TensorFlow.js platform patch:', error);
|
||||
}
|
||||
}
|
||||
|
||||
const fn = new Function('return ' + workerData.fnString)();
|
||||
const result = fn(workerData.args);
|
||||
parentPort.postMessage({ result });
|
||||
`,
|
||||
{
|
||||
eval: true,
|
||||
workerData: { fnString, args }
|
||||
}
|
||||
)
|
||||
workerPool.set(randomKey, worker)
|
||||
}
|
||||
|
||||
worker._busy = true
|
||||
}
|
||||
|
||||
worker.on('message', (message: any) => {
|
||||
worker._busy = false
|
||||
resolve(message.result as T)
|
||||
})
|
||||
|
||||
worker.on('error', (err: any) => {
|
||||
worker._busy = false
|
||||
reject(err)
|
||||
})
|
||||
|
||||
worker.on('exit', (code: number) => {
|
||||
if (code !== 0) {
|
||||
worker._busy = false
|
||||
reject(new Error(`Worker stopped with exit code ${code}`))
|
||||
}
|
||||
})
|
||||
})
|
||||
.catch(reject)
|
||||
} catch (error) {
|
||||
reject(error)
|
||||
}
|
||||
})
|
||||
}
|
||||
|
||||
/**
|
||||
* Execute a function in a Web Worker (Browser environment)
|
||||
*/
|
||||
function executeInWebWorker<T>(fnString: string, args: any): Promise<T> {
|
||||
return new Promise<T>((resolve, reject) => {
|
||||
try {
|
||||
// Use the dedicated worker.js file instead of creating a blob
|
||||
// Try different approaches to locate the worker.js file
|
||||
let workerPath = './worker.js'
|
||||
|
||||
try {
|
||||
// First try to use the import.meta.url if available (modern browsers)
|
||||
if (typeof import.meta !== 'undefined' && import.meta.url) {
|
||||
const baseUrl = import.meta.url.substring(
|
||||
0,
|
||||
import.meta.url.lastIndexOf('/') + 1
|
||||
)
|
||||
workerPath = `${baseUrl}worker.js`
|
||||
}
|
||||
// Fallback to a relative path based on the unified.js location
|
||||
else if (typeof document !== 'undefined') {
|
||||
// Find the script tag that loaded unified.js
|
||||
const scripts = document.getElementsByTagName('script')
|
||||
for (let i = 0; i < scripts.length; i++) {
|
||||
const src = scripts[i].src
|
||||
if (src && src.includes('unified.js')) {
|
||||
// Get the directory path
|
||||
workerPath =
|
||||
src.substring(0, src.lastIndexOf('/') + 1) + 'worker.js'
|
||||
break
|
||||
}
|
||||
}
|
||||
}
|
||||
} catch (e) {
|
||||
console.warn(
|
||||
'Could not determine worker path from import.meta.url, using relative path',
|
||||
e
|
||||
)
|
||||
}
|
||||
|
||||
// If we couldn't determine the path, try some common locations
|
||||
if (workerPath === './worker.js' && typeof window !== 'undefined') {
|
||||
// Try to find the worker.js in the same directory as the current page
|
||||
const pageUrl = window.location.href
|
||||
const pageDir = pageUrl.substring(0, pageUrl.lastIndexOf('/') + 1)
|
||||
workerPath = `${pageDir}worker.js`
|
||||
|
||||
// Also check for dist/worker.js
|
||||
if (typeof document !== 'undefined') {
|
||||
const distWorkerPath = `${pageDir}dist/worker.js`
|
||||
// Create a test request to see if the file exists
|
||||
const xhr = new XMLHttpRequest()
|
||||
xhr.open('HEAD', distWorkerPath, false)
|
||||
try {
|
||||
xhr.send()
|
||||
if (xhr.status >= 200 && xhr.status < 300) {
|
||||
workerPath = distWorkerPath
|
||||
}
|
||||
} catch (e) {
|
||||
// Ignore errors, we'll use the default path
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
console.log('Using worker path:', workerPath)
|
||||
|
||||
// Try to create a worker, but fall back to inline worker or main thread execution if it fails
|
||||
let worker: Worker
|
||||
try {
|
||||
worker = new Worker(workerPath)
|
||||
} catch (error) {
|
||||
console.warn(
|
||||
'Failed to create Web Worker from file, trying inline worker:',
|
||||
error
|
||||
)
|
||||
|
||||
try {
|
||||
// Create an inline worker using a Blob
|
||||
const workerCode = `
|
||||
// Brainy Inline Worker Script
|
||||
console.log('Brainy Inline Worker: Started');
|
||||
|
||||
self.onmessage = function (e) {
|
||||
try {
|
||||
console.log('Brainy Inline Worker: Received message', e.data ? 'with data' : 'without data');
|
||||
|
||||
if (!e.data || !e.data.fnString) {
|
||||
throw new Error('Invalid message: missing function string');
|
||||
}
|
||||
|
||||
console.log('Brainy Inline Worker: Creating function from string');
|
||||
const fn = new Function('return ' + e.data.fnString)();
|
||||
|
||||
console.log('Brainy Inline Worker: Executing function with args');
|
||||
const result = fn(e.data.args);
|
||||
|
||||
console.log('Brainy Inline Worker: Function executed successfully, posting result');
|
||||
self.postMessage({ result: result });
|
||||
} catch (error) {
|
||||
console.error('Brainy Inline Worker: Error executing function', error);
|
||||
self.postMessage({
|
||||
error: error.message,
|
||||
stack: error.stack
|
||||
});
|
||||
}
|
||||
};
|
||||
`
|
||||
|
||||
const blob = new Blob([workerCode], {
|
||||
type: 'application/javascript'
|
||||
})
|
||||
const blobUrl = URL.createObjectURL(blob)
|
||||
worker = new Worker(blobUrl)
|
||||
|
||||
console.log('Created inline worker using Blob URL')
|
||||
} catch (inlineWorkerError) {
|
||||
console.warn(
|
||||
'Failed to create inline Web Worker, falling back to main thread execution:',
|
||||
inlineWorkerError
|
||||
)
|
||||
// Execute in main thread as fallback
|
||||
try {
|
||||
const fn = new Function('return ' + fnString)()
|
||||
resolve(fn(args) as T)
|
||||
return
|
||||
} catch (mainThreadError) {
|
||||
reject(mainThreadError)
|
||||
return
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// Set a timeout to prevent hanging
|
||||
const timeoutId = setTimeout(() => {
|
||||
console.warn(
|
||||
'Web Worker execution timed out, falling back to main thread'
|
||||
)
|
||||
worker.terminate()
|
||||
|
||||
// Execute in main thread as fallback
|
||||
try {
|
||||
const fn = new Function('return ' + fnString)()
|
||||
resolve(fn(args) as T)
|
||||
} catch (mainThreadError) {
|
||||
reject(mainThreadError)
|
||||
}
|
||||
}, 25000) // 25 second timeout (less than the 30 second test timeout)
|
||||
|
||||
worker.onmessage = function (e) {
|
||||
clearTimeout(timeoutId)
|
||||
if (e.data.error) {
|
||||
reject(new Error(e.data.error))
|
||||
} else {
|
||||
resolve(e.data.result as T)
|
||||
}
|
||||
worker.terminate()
|
||||
}
|
||||
|
||||
worker.onerror = function (e) {
|
||||
clearTimeout(timeoutId)
|
||||
console.warn(
|
||||
'Web Worker error, falling back to main thread execution:',
|
||||
e.message
|
||||
)
|
||||
worker.terminate()
|
||||
|
||||
// Execute in main thread as fallback
|
||||
try {
|
||||
const fn = new Function('return ' + fnString)()
|
||||
resolve(fn(args) as T)
|
||||
} catch (mainThreadError) {
|
||||
reject(mainThreadError)
|
||||
}
|
||||
}
|
||||
|
||||
worker.postMessage({ fnString, args })
|
||||
} catch (error) {
|
||||
reject(error)
|
||||
}
|
||||
})
|
||||
}
|
||||
|
||||
/**
|
||||
* Clean up all worker pools
|
||||
* This should be called when the application is shutting down
|
||||
*/
|
||||
export function cleanupWorkerPools(): void {
|
||||
if (isNode()) {
|
||||
import('node:worker_threads')
|
||||
.then(({ Worker }) => {
|
||||
for (const worker of workerPool.values()) {
|
||||
worker.terminate()
|
||||
}
|
||||
workerPool.clear()
|
||||
console.log('Worker pools cleaned up')
|
||||
})
|
||||
.catch(console.error)
|
||||
}
|
||||
}
|
||||
|
|
@ -1,72 +0,0 @@
|
|||
// Brainy Worker Script
|
||||
// This script is used by the workerUtils.js file to execute functions in a separate thread
|
||||
|
||||
// Note: TensorFlow.js platform patch is applied in setup.ts
|
||||
// Worker scripts should import setup.ts if they need TensorFlow.js functionality
|
||||
|
||||
// Log that the worker has started
|
||||
console.log('Brainy Worker: Started')
|
||||
|
||||
// Define the message handler with proper TypeScript typing
|
||||
self.onmessage = function (e: MessageEvent): void {
|
||||
try {
|
||||
console.log(
|
||||
'Brainy Worker: Received message',
|
||||
e.data ? 'with data' : 'without data'
|
||||
)
|
||||
|
||||
if (!e.data || !e.data.fnString) {
|
||||
throw new Error('Invalid message: missing function string')
|
||||
}
|
||||
|
||||
console.log('Brainy Worker: Creating function from string')
|
||||
// Use Function constructor to create a function from the string
|
||||
let fn
|
||||
|
||||
try {
|
||||
// First try with 'return' prefix
|
||||
fn = new Function('return ' + e.data.fnString)()
|
||||
} catch (functionError) {
|
||||
console.warn(
|
||||
'Brainy Worker: Error creating function with return syntax, trying alternative approaches',
|
||||
functionError
|
||||
)
|
||||
|
||||
try {
|
||||
// Try wrapping in parentheses for function expressions
|
||||
fn = new Function('return (' + e.data.fnString + ')')()
|
||||
} catch (wrapError) {
|
||||
console.warn(
|
||||
'Brainy Worker: Error creating function with parentheses wrapping',
|
||||
wrapError
|
||||
)
|
||||
|
||||
try {
|
||||
// Try direct approach for named functions
|
||||
fn = new Function(e.data.fnString)()
|
||||
} catch (directError) {
|
||||
console.error(
|
||||
'Brainy Worker: All approaches to create function failed',
|
||||
directError
|
||||
)
|
||||
throw new Error(
|
||||
'Failed to create function from string: ' +
|
||||
(functionError as Error).message
|
||||
)
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
console.log('Brainy Worker: Executing function with args')
|
||||
const result = fn(e.data.args)
|
||||
|
||||
console.log('Brainy Worker: Function executed successfully, posting result')
|
||||
self.postMessage({ result: result })
|
||||
} catch (error: any) {
|
||||
console.error('Brainy Worker: Error executing function', error)
|
||||
self.postMessage({
|
||||
error: error.message,
|
||||
stack: error.stack
|
||||
})
|
||||
}
|
||||
}
|
||||
Loading…
Add table
Add a link
Reference in a new issue