LibreChat/api/server/services/Endpoints/agents/initialize.js
Danny Avila 37964975c1
🤖 refactor: Improve Agents Memory Usage, Bump Keyv, Grok 3 (#6850)
* chore: remove unused redis file

* chore: bump keyv dependencies, and update related imports

* refactor: Implement IoRedis client for rate limiting across middleware, as node-redis via keyv not compatible

* fix: Set max listeners to expected amount

* WIP: memory improvements

* refactor: Simplify getAbortData assignment in createAbortController

* refactor: Update getAbortData to use WeakRef for content management

* WIP: memory improvements in agent chat requests

* refactor: Enhance memory management with finalization registry and cleanup functions

* refactor: Simplify domainParser calls by removing unnecessary request parameter

* refactor: Update parameter types for action tools and agent loading functions to use minimal configs

* refactor: Simplify domainParser tests by removing unnecessary request parameter

* refactor: Simplify domainParser call by removing unnecessary request parameter

* refactor: Enhance client disposal by nullifying additional properties to improve memory management

* refactor: Improve title generation by adding abort controller and timeout handling, consolidate request cleanup

* refactor: Update checkIdleConnections to skip current user when checking for idle connections if passed

* refactor: Update createMCPTool to derive userId from config and handle abort signals

* refactor: Introduce createTokenCounter function and update tokenCounter usage; enhance disposeClient to reset Graph values

* refactor: Update getMCPManager to accept userId parameter for improved idle connection handling

* refactor: Extract logToolError function for improved error handling in AgentClient

* refactor: Update disposeClient to clear handlerRegistry and graphRunnable references in client.run

* refactor: Extract createHandleNewToken function to streamline token handling in initializeClient

* chore: bump @librechat/agents

* refactor: Improve timeout handling in addTitle function for better error management

* refactor: Introduce createFetch instead of using class method

* refactor: Enhance client disposal and request data handling in AskController and EditController

* refactor: Update import statements for AnthropicClient and OpenAIClient to use specific paths

* refactor: Use WeakRef for response handling in SplitStreamHandler to prevent memory leaks

* refactor: Simplify client disposal and rename getReqData to processReqData in AskController and EditController

* refactor: Improve logging structure and parameter handling in OpenAIClient

* refactor: Remove unused GraphEvents and improve stream event handling in AnthropicClient and OpenAIClient

* refactor: Simplify client initialization in AskController and EditController

* refactor: Remove unused mock functions and implement in-memory store for KeyvMongo

* chore: Update dependencies in package-lock.json to latest versions

* refactor: Await token usage recording in OpenAIClient to ensure proper async handling

* refactor: Remove handleAbort route from multiple endpoints and enhance client disposal logic

* refactor: Enhance abort controller logic by managing abortKey more effectively

* refactor: Add newConversation handling in useEventHandlers for improved conversation management

* fix: dropparams

* refactor: Use optional chaining for safer access to request properties in BaseClient

* refactor: Move client disposal and request data processing logic to cleanup module for better organization

* refactor: Remove aborted request check from addTitle function for cleaner logic

* feat: Add Grok 3 model pricing and update tests for new models

* chore: Remove trace warnings and inspect flags from backend start script used for debugging

* refactor: Replace user identifier handling with userId for consistency across controllers, use UserId in clientRegistry

* refactor: Enhance client disposal logic to prevent memory leaks by clearing additional references

* chore: Update @librechat/agents to version 2.4.14 in package.json and package-lock.json
2025-04-12 18:46:36 -04:00

345 lines
10 KiB
JavaScript

const { createContentAggregator, Providers } = require('@librechat/agents');
const {
Constants,
ErrorTypes,
EModelEndpoint,
getResponseSender,
AgentCapabilities,
providerEndpointMap,
} = require('librechat-data-provider');
const {
getDefaultHandlers,
createToolEndCallback,
} = require('~/server/controllers/agents/callbacks');
const initAnthropic = require('~/server/services/Endpoints/anthropic/initialize');
const getBedrockOptions = require('~/server/services/Endpoints/bedrock/options');
const initOpenAI = require('~/server/services/Endpoints/openAI/initialize');
const initCustom = require('~/server/services/Endpoints/custom/initialize');
const initGoogle = require('~/server/services/Endpoints/google/initialize');
const generateArtifactsPrompt = require('~/app/clients/prompts/artifacts');
const { getCustomEndpointConfig } = require('~/server/services/Config');
const { processFiles } = require('~/server/services/Files/process');
const { loadAgentTools } = require('~/server/services/ToolService');
const AgentClient = require('~/server/controllers/agents/client');
const { getConvoFiles } = require('~/models/Conversation');
const { getToolFilesByIds } = require('~/models/File');
const { getModelMaxTokens } = require('~/utils');
const { getAgent } = require('~/models/Agent');
const { getFiles } = require('~/models/File');
const { logger } = require('~/config');
const providerConfigMap = {
[Providers.XAI]: initCustom,
[Providers.OLLAMA]: initCustom,
[Providers.DEEPSEEK]: initCustom,
[Providers.OPENROUTER]: initCustom,
[EModelEndpoint.openAI]: initOpenAI,
[EModelEndpoint.google]: initGoogle,
[EModelEndpoint.azureOpenAI]: initOpenAI,
[EModelEndpoint.anthropic]: initAnthropic,
[EModelEndpoint.bedrock]: getBedrockOptions,
};
/**
* @param {ServerRequest} req
* @param {Promise<Array<MongoFile | null>> | undefined} _attachments
* @param {AgentToolResources | undefined} _tool_resources
* @returns {Promise<{ attachments: Array<MongoFile | undefined> | undefined, tool_resources: AgentToolResources | undefined }>}
*/
const primeResources = async (req, _attachments, _tool_resources) => {
try {
/** @type {Array<MongoFile | undefined> | undefined} */
let attachments;
const tool_resources = _tool_resources ?? {};
const isOCREnabled = (req.app.locals?.[EModelEndpoint.agents]?.capabilities ?? []).includes(
AgentCapabilities.ocr,
);
if (tool_resources.ocr?.file_ids && isOCREnabled) {
const context = await getFiles(
{
file_id: { $in: tool_resources.ocr.file_ids },
},
{},
{},
);
attachments = (attachments ?? []).concat(context);
}
if (!_attachments) {
return { attachments, tool_resources };
}
/** @type {Array<MongoFile | undefined> | undefined} */
const files = await _attachments;
if (!attachments) {
/** @type {Array<MongoFile | undefined>} */
attachments = [];
}
for (const file of files) {
if (!file) {
continue;
}
if (file.metadata?.fileIdentifier) {
const execute_code = tool_resources.execute_code ?? {};
if (!execute_code.files) {
tool_resources.execute_code = { ...execute_code, files: [] };
}
tool_resources.execute_code.files.push(file);
} else if (file.embedded === true) {
const file_search = tool_resources.file_search ?? {};
if (!file_search.files) {
tool_resources.file_search = { ...file_search, files: [] };
}
tool_resources.file_search.files.push(file);
}
attachments.push(file);
}
return { attachments, tool_resources };
} catch (error) {
logger.error('Error priming resources', error);
return { attachments: _attachments, tool_resources: _tool_resources };
}
};
/**
* @param {...string | number} values
* @returns {string | number | undefined}
*/
function optionalChainWithEmptyCheck(...values) {
for (const value of values) {
if (value !== undefined && value !== null && value !== '') {
return value;
}
}
return values[values.length - 1];
}
/**
* @param {object} params
* @param {ServerRequest} params.req
* @param {ServerResponse} params.res
* @param {Agent} params.agent
* @param {Set<string>} [params.allowedProviders]
* @param {object} [params.endpointOption]
* @param {boolean} [params.isInitialAgent]
* @returns {Promise<Agent>}
*/
const initializeAgentOptions = async ({
req,
res,
agent,
endpointOption,
allowedProviders,
isInitialAgent = false,
}) => {
if (allowedProviders.size > 0 && !allowedProviders.has(agent.provider)) {
throw new Error(
`{ "type": "${ErrorTypes.INVALID_AGENT_PROVIDER}", "info": "${agent.provider}" }`,
);
}
let currentFiles;
/** @type {Array<MongoFile>} */
const requestFiles = req.body.files ?? [];
if (
isInitialAgent &&
req.body.conversationId != null &&
(agent.model_parameters?.resendFiles ?? true) === true
) {
const fileIds = (await getConvoFiles(req.body.conversationId)) ?? [];
const toolFiles = await getToolFilesByIds(fileIds);
if (requestFiles.length || toolFiles.length) {
currentFiles = await processFiles(requestFiles.concat(toolFiles));
}
} else if (isInitialAgent && requestFiles.length) {
currentFiles = await processFiles(requestFiles);
}
const { attachments, tool_resources } = await primeResources(
req,
currentFiles,
agent.tool_resources,
);
const provider = agent.provider;
const { tools, toolContextMap } = await loadAgentTools({
req,
res,
agent: {
id: agent.id,
tools: agent.tools,
provider,
model: agent.model,
},
tool_resources,
});
agent.endpoint = provider;
let getOptions = providerConfigMap[provider];
if (!getOptions && providerConfigMap[provider.toLowerCase()] != null) {
agent.provider = provider.toLowerCase();
getOptions = providerConfigMap[agent.provider];
} else if (!getOptions) {
const customEndpointConfig = await getCustomEndpointConfig(provider);
if (!customEndpointConfig) {
throw new Error(`Provider ${provider} not supported`);
}
getOptions = initCustom;
agent.provider = Providers.OPENAI;
}
const model_parameters = Object.assign(
{},
agent.model_parameters ?? { model: agent.model },
isInitialAgent === true ? endpointOption?.model_parameters : {},
);
const _endpointOption =
isInitialAgent === true
? Object.assign({}, endpointOption, { model_parameters })
: { model_parameters };
const options = await getOptions({
req,
res,
optionsOnly: true,
overrideEndpoint: provider,
overrideModel: agent.model,
endpointOption: _endpointOption,
});
if (options.provider != null) {
agent.provider = options.provider;
}
/** @type {import('@librechat/agents').ClientOptions} */
agent.model_parameters = Object.assign(model_parameters, options.llmConfig);
if (options.configOptions) {
agent.model_parameters.configuration = options.configOptions;
}
if (!agent.model_parameters.model) {
agent.model_parameters.model = agent.model;
}
if (typeof agent.artifacts === 'string' && agent.artifacts !== '') {
agent.additional_instructions = generateArtifactsPrompt({
endpoint: agent.provider,
artifacts: agent.artifacts,
});
}
const tokensModel =
agent.provider === EModelEndpoint.azureOpenAI ? agent.model : agent.model_parameters.model;
const maxTokens = optionalChainWithEmptyCheck(
agent.model_parameters.maxOutputTokens,
agent.model_parameters.maxTokens,
0,
);
const maxContextTokens = optionalChainWithEmptyCheck(
agent.model_parameters.maxContextTokens,
agent.max_context_tokens,
getModelMaxTokens(tokensModel, providerEndpointMap[provider]),
4096,
);
return {
...agent,
tools,
attachments,
toolContextMap,
maxContextTokens: (maxContextTokens - maxTokens) * 0.9,
};
};
const initializeClient = async ({ req, res, endpointOption }) => {
if (!endpointOption) {
throw new Error('Endpoint option not provided');
}
// TODO: use endpointOption to determine options/modelOptions
/** @type {Array<UsageMetadata>} */
const collectedUsage = [];
/** @type {ArtifactPromises} */
const artifactPromises = [];
const { contentParts, aggregateContent } = createContentAggregator();
const toolEndCallback = createToolEndCallback({ req, res, artifactPromises });
const eventHandlers = getDefaultHandlers({
res,
aggregateContent,
toolEndCallback,
collectedUsage,
});
if (!endpointOption.agent) {
throw new Error('No agent promise provided');
}
// Initialize primary agent
const primaryAgent = await endpointOption.agent;
if (!primaryAgent) {
throw new Error('Agent not found');
}
const agentConfigs = new Map();
/** @type {Set<string>} */
const allowedProviders = new Set(req?.app?.locals?.[EModelEndpoint.agents]?.allowedProviders);
// Handle primary agent
const primaryConfig = await initializeAgentOptions({
req,
res,
agent: primaryAgent,
endpointOption,
allowedProviders,
isInitialAgent: true,
});
const agent_ids = primaryConfig.agent_ids;
if (agent_ids?.length) {
for (const agentId of agent_ids) {
const agent = await getAgent({ id: agentId });
if (!agent) {
throw new Error(`Agent ${agentId} not found`);
}
const config = await initializeAgentOptions({
req,
res,
agent,
endpointOption,
allowedProviders,
});
agentConfigs.set(agentId, config);
}
}
const sender =
primaryAgent.name ??
getResponseSender({
...endpointOption,
model: endpointOption.model_parameters.model,
});
const client = new AgentClient({
req,
res,
sender,
contentParts,
agentConfigs,
eventHandlers,
collectedUsage,
aggregateContent,
artifactPromises,
agent: primaryConfig,
spec: endpointOption.spec,
iconURL: endpointOption.iconURL,
attachments: primaryConfig.attachments,
endpointType: endpointOption.endpointType,
maxContextTokens: primaryConfig.maxContextTokens,
resendFiles: primaryConfig.model_parameters?.resendFiles ?? true,
endpoint:
primaryConfig.id === Constants.EPHEMERAL_AGENT_ID
? primaryConfig.endpoint
: EModelEndpoint.agents,
});
return { client };
};
module.exports = { initializeClient };