🔎 feat: Native Web Search with Citation References (#7516)

* WIP: search tool integration

* WIP: Add web search capabilities and API key management to agent actions

* WIP: web search capability to agent configuration and selection

* WIP: Add web search capability to backend agent configuration

* WIP: add web search option to default agent form values

* WIP: add attachments for web search

* feat: add plugin for processing web search citations

* WIP: first pass, Citation UI

* chore: remove console.log

* feat: Add AnimatedTabs component for tabbed UI functionality

* refactor: AnimatedTabs component with CSS animations and stable ID generation

* WIP example content

* feat: SearchContext for managing search results apart from MessageContext

* feat: Enhance AnimatedTabs with underline animation and state management

* WIP: first pass, Implement dynamic tab functionality in Sources component with search results integration

* fix: Update class names for improved styling in Sources and AnimatedTabs components

* feat: Improve styling and layout in Sources component with enhanced button and item designs

* feat: Refactor Sources component to integrate OGDialog for source display and improve layout

* style: Update background color in SourceItem and SourcesGroup components for improved visibility

* refactor: Sources component to enhance SourceItem structure and improve favicon handling

* style: Adjust font size of domain text in SourceItem for better readability

* feat: Add localization for citation source and details in CompositeCitation component

* style: add theming to Citation components

* feat: Enhance SourceItem component with dialog support and improved hovercard functionality

* feat: Add localization for sources tab and image alt text in Sources component

* style: Replace divs with spans for better semantic structure in CompositeCitation and Citation components

* refactor: Sources component to use useMemo for tab generation and improve performance

* chore: bump @librechat/agents to v2.4.318

* chore: update search result types

* fix: search results retrieval in ContentParts component, re-render attachments when expected

* feat: update sources style/types to use latest search result structure

* style: enhance Dialog (expanded) SourceItem component with link wrapping and improved styling

* style: update ImageItem component styling for improved title visibility

* refactor: remove SourceItemBase component and adjust SourceItem layout for improved styling

* chore: linting twcss order

* fix: prevent FileAttachment from rendering search attachments

* fix: append underscore to responseMessageId for unique identification to prevent mapping of previous latest message's attachments

* chore: remove unused parameter 'useSpecs' from loadTools function

* chore: twcss order

* WIP: WebSearch Tool UI

* refactor: add limit parameter to StackedFavicons for customizable source display

* refactor: optimize search results memoization by making more granular and separate conerns

* refactor: integrated StackedFavicons to WebSearch mid-run

* chore: bump @librechat/agents to expose handleToolCallChunks

* chore: use typedefs from dedicated file instead of defining them in AgentClient module

* WIP: first pass, search progress results

* refactor: move createOnSearchResults function to a dedicated search module

* chore: bump @librechat/agents to v2.4.320

* WIP: first pass, search results processed UX

* refactor: consolidate context variables in createOnSearchResults function

* chore: bump @librechat/agents to v2.4.321

* feat: add guidelines for web search tool response formatting in loadTools function

* feat: add isLast prop to Part component and update WebSearch logic for improved state handling

* style: update Hovercard styles for improved UI consistency

* feat: export FaviconImage component for improved accessibility in other modules

* refactor: export getCleanDomain function and use FaviconImage in Citation component for improved source representation

* refactor: implement SourceHovercard component for consistency and DRY compliance

* fix: replace <p> with <span> for snippet and title in SourceItem and SourceHovercard for consistency

* style: `not-prose`

* style: remove 'not-prose' class for consistency in SourceItem, Citation, and SourceHovercard components, adjust style classes

* refactor: `imageUrl` on hover and prevent duplicate sources

* refactor: enhance SourcesGroup dialog layout and improve source item presentation

* refactor: reorganize Web Components, save in same directory

* feat: add 'news' refType to refTypeMap for citation sources

* style: adjust Hovercard width for improved layout

* refactor: update tool usage guidelines for improved clarity and execution

* chore: linting

* feat: add Web Search badge with initial permissions and local storage logic

* feat: add webSearch support to interface and permissions schemas

* feat: implement Web Search API key management and localization updates

* feat: refactor Web Search API key handling and integrate new search API key form

* fix: remove unnecessary visibility state from FileAttachment component

* feat: update WebSearch component to use Globe icon and localized search label

* feat: enhance ApiKeyDialog with dropdown for reranker selection and update translations

* feat: implement dropdown menus for engine, scraper, and reranker selection in ApiKeyDialog

* chore: linting and add unknown instead of `any` type

* feat: refactor ApiKeyDialog and useAuthSearchTool for improved API key management

* refactor: update ocrSchema to use template literals for default apiKey and baseURL

* feat: add web search configuration and utility functions for environment variable extraction

* fix: ensure filepath is defined before checking its prefix in useAttachmentHandler

* feat: enhance web search functionality with improved configuration and environment variable extraction for authFields

* fix: update auth type in TPluginAction and TUpdateUserPlugins to use Partial<Record<string, string>>

* feat: implement web search authentication verification and enhance webSearchAuth structure

* feat: enhance ephemeral agent handling with new web search capability and type definition

* feat: enhance isEphemeralAgent function to include web search selection

* feat: refactor verifyWebSearchAuth to improve key handling and authentication checks

* feat: implement loadWebSearchAuth function for improved web search authentication handling

* feat: enhance web search authentication with new configuration options and refactor related types

* refactor: rename search engine to search provider and update related localization keys

* feat: update verifyWebSearchAuth to handle multiple authentication types and improve error handling

* feat: update ApiKeyDialog to accept authTypes prop and remove isUserProvided check

* feat: add tests for extractWebSearchEnvVars and loadWebSearchAuth functions

* feat: enhance loadWebSearchAuth to support specific service checks for providers, scrapers, and rerankers

* fix: update web search configuration key and adjust auth result handling in loadTools function

* feat: add new progress key for repeated web searching and update localization

* chore: bump @librechat/agents to 2.4.322

* feat: enhance loadTools function to include ISO time and improve search tool logging

* feat: update StackedFavicons to handle negative start index and improve citation attribution styling and text

* chore: update .gitignore to categorize AI-related files

* fix: mobile responsiveness of sources/citations hovercards

* feat: enhance source display with improved line clamping for better readability

* chore: bump @librechat/agents to v2.4.33

* feat: add handling for image sources in references mapping

* chore: bump librechat-data-provider version to 0.7.84

* chore: bump @librechat/agents version to 2.4.34

* fix: update auth handling to support multiple auth types in tools and allow key configuration in agent panel

* chore: remove redundant agent attribution text from search form

* fix: web search auth uninstall

* refactor: convert CheckboxButton to a forwardRef component and update setValue callback signature

* feat: add triggerRef prop to ApiKeyDialog components for improved dialog control

* feat: integrate triggerRef in CodeInterpreter and WebSearch components for enhanced dialog management

* feat: enhance ApiKeyDialog with additional links for Firecrawl and Jina API key guidance

* feat: implement web search configuration handling in ApiKeyDialog and add tests for dropdown visibility

* fix: update webSearchConfig reference in config route for correct payload assignment

* feat: update ApiKeyDialog to conditionally render sections based on authTypes and modify loadWebSearchAuth to correctly categorize authentication types

* feat: refactor ApiKeyDialog and related tests to use SearchCategories and RerankerTypes enums and remove nested ternaries

* refactor: move ThinkingButton rendering to improve layout consistency in ContentParts

* feat: integrate search context into Markdown component to conditionally include unicodeCitation plugin

* chore: bump @librechat/agents to v2.4.35

* chore: remove unused 18n key

* ci: add WEB_SEARCH permission testing and update AppService tests for new webSearch configuration

* ci: add more comprehensive tests for loadWebSearchAuth to validate authentication handling and authTypes structure

* chore: remove debugging console log from web.spec.ts to clean up test output
This commit is contained in:
Danny Avila 2025-05-23 00:14:04 -04:00
parent bf80cf30b3
commit 0dbbf7de04
No known key found for this signature in database
GPG key ID: BF31EEB2C5CA0956
73 changed files with 6366 additions and 2003 deletions

View file

@ -0,0 +1,899 @@
import type {
ScraperTypes,
TCustomConfig,
RerankerTypes,
SearchProviders,
TWebSearchConfig,
} from '../src/config';
import { webSearchAuth, loadWebSearchAuth, extractWebSearchEnvVars } from '../src/web';
import { AuthType } from '../src/schemas';
// Mock the extractVariableName function
jest.mock('../src/utils', () => ({
extractVariableName: (value: string) => {
if (!value || typeof value !== 'string') return null;
const match = value.match(/^\${(.+)}$/);
return match ? match[1] : null;
},
}));
describe('web.ts', () => {
describe('extractWebSearchEnvVars', () => {
it('should return empty array if config is undefined', () => {
const result = extractWebSearchEnvVars({
keys: ['serperApiKey', 'jinaApiKey'],
config: undefined,
});
expect(result).toEqual([]);
});
it('should extract environment variable names from config values', () => {
const config: Partial<TWebSearchConfig> = {
serperApiKey: '${SERPER_API_KEY}',
jinaApiKey: '${JINA_API_KEY}',
cohereApiKey: 'actual-api-key', // Not in env var format
safeSearch: true,
};
const result = extractWebSearchEnvVars({
keys: ['serperApiKey', 'jinaApiKey', 'cohereApiKey'],
config: config as TWebSearchConfig,
});
expect(result).toEqual(['SERPER_API_KEY', 'JINA_API_KEY']);
});
it('should only extract variables for keys that exist in the config', () => {
const config: Partial<TWebSearchConfig> = {
serperApiKey: '${SERPER_API_KEY}',
// firecrawlApiKey is missing
safeSearch: true,
};
const result = extractWebSearchEnvVars({
keys: ['serperApiKey', 'firecrawlApiKey'],
config: config as TWebSearchConfig,
});
expect(result).toEqual(['SERPER_API_KEY']);
});
});
describe('loadWebSearchAuth', () => {
// Common test variables
const userId = 'test-user-id';
let mockLoadAuthValues: jest.Mock;
let webSearchConfig: TCustomConfig['webSearch'];
beforeEach(() => {
// Reset mocks before each test
jest.clearAllMocks();
// Initialize the mock function
mockLoadAuthValues = jest.fn();
// Initialize a basic webSearchConfig
webSearchConfig = {
serperApiKey: '${SERPER_API_KEY}',
firecrawlApiKey: '${FIRECRAWL_API_KEY}',
firecrawlApiUrl: '${FIRECRAWL_API_URL}',
jinaApiKey: '${JINA_API_KEY}',
cohereApiKey: '${COHERE_API_KEY}',
safeSearch: true,
};
});
it('should return authenticated=true when all required categories are authenticated', async () => {
// Mock successful authentication for all services
mockLoadAuthValues.mockImplementation(({ authFields }) => {
const result: Record<string, string> = {};
authFields.forEach((field) => {
result[field] =
field === 'FIRECRAWL_API_URL' ? 'https://api.firecrawl.dev' : 'test-api-key';
});
return Promise.resolve(result);
});
const result = await loadWebSearchAuth({
userId,
webSearchConfig,
loadAuthValues: mockLoadAuthValues,
});
expect(result.authenticated).toBe(true);
expect(result.authTypes).toHaveLength(3); // providers, scrapers, rerankers
expect(result.authResult).toHaveProperty('serperApiKey', 'test-api-key');
expect(result.authResult).toHaveProperty('firecrawlApiKey', 'test-api-key');
// The implementation only includes one reranker in the result
// It will be either jina or cohere, but not both
if (result.authResult.rerankerType === 'jina') {
expect(result.authResult).toHaveProperty('jinaApiKey', 'test-api-key');
} else {
expect(result.authResult).toHaveProperty('cohereApiKey', 'test-api-key');
}
expect(result.authResult).toHaveProperty('searchProvider', 'serper');
expect(result.authResult).toHaveProperty('scraperType', 'firecrawl');
expect(['jina', 'cohere']).toContain(result.authResult.rerankerType as string);
});
it('should return authenticated=false when a required category is not authenticated', async () => {
// Mock authentication failure for the providers category
mockLoadAuthValues.mockImplementation(({ authFields }) => {
const result: Record<string, string> = {};
authFields.forEach((field) => {
// Only provide values for scrapers and rerankers, not for providers
if (field !== 'SERPER_API_KEY') {
result[field] =
field === 'FIRECRAWL_API_URL' ? 'https://api.firecrawl.dev' : 'test-api-key';
}
});
return Promise.resolve(result);
});
const result = await loadWebSearchAuth({
userId,
webSearchConfig,
loadAuthValues: mockLoadAuthValues,
});
expect(result.authenticated).toBe(false);
// We should still have authTypes for the categories we checked
expect(result.authTypes.some(([category]) => category === 'providers')).toBe(true);
});
it('should handle exceptions from loadAuthValues', async () => {
// Mock loadAuthValues to throw an error
mockLoadAuthValues.mockImplementation(() => {
throw new Error('Authentication failed');
});
const result = await loadWebSearchAuth({
userId,
webSearchConfig,
loadAuthValues: mockLoadAuthValues,
throwError: false, // Don't throw errors
});
expect(result.authenticated).toBe(false);
});
it('should correctly identify user-provided vs system-defined auth', async () => {
// Mock environment variables
const originalEnv = process.env;
process.env = {
...originalEnv,
SERPER_API_KEY: 'system-api-key',
FIRECRAWL_API_KEY: 'system-api-key',
JINA_API_KEY: 'system-api-key',
};
// Mock loadAuthValues to return different values for some keys
mockLoadAuthValues.mockImplementation(({ authFields }) => {
const result: Record<string, string> = {};
authFields.forEach((field) => {
if (field === 'SERPER_API_KEY') {
// This matches the system env var
result[field] = 'system-api-key';
} else if (field === 'FIRECRAWL_API_KEY') {
// This is different from the system env var (user provided)
result[field] = 'user-api-key';
} else if (field === 'FIRECRAWL_API_URL') {
result[field] = 'https://api.firecrawl.dev';
} else if (field === 'JINA_API_KEY') {
// This matches the system env var
result[field] = 'system-api-key';
} else {
result[field] = 'test-api-key';
}
});
return Promise.resolve(result);
});
const result = await loadWebSearchAuth({
userId,
webSearchConfig,
loadAuthValues: mockLoadAuthValues,
});
expect(result.authenticated).toBe(true);
// Check for providers (system-defined) and scrapers (user-provided)
const providersAuthType = result.authTypes.find(
([category]) => category === 'providers',
)?.[1];
const scrapersAuthType = result.authTypes.find(([category]) => category === 'scrapers')?.[1];
expect(providersAuthType).toBe(AuthType.SYSTEM_DEFINED);
expect(scrapersAuthType).toBe(AuthType.USER_PROVIDED);
// Restore original env
process.env = originalEnv;
});
it('should handle optional fields correctly', async () => {
// Create a config without the optional firecrawlApiUrl
const configWithoutOptional = { ...webSearchConfig } as Partial<TWebSearchConfig>;
delete configWithoutOptional.firecrawlApiUrl;
mockLoadAuthValues.mockImplementation(({ authFields, optional }) => {
const result: Record<string, string> = {};
authFields.forEach((field) => {
// Don't provide values for optional fields
if (!optional?.has(field)) {
result[field] = 'test-api-key';
}
});
return Promise.resolve(result);
});
const result = await loadWebSearchAuth({
userId,
webSearchConfig: configWithoutOptional as TWebSearchConfig,
loadAuthValues: mockLoadAuthValues,
});
expect(result.authenticated).toBe(true);
expect(result.authResult).toHaveProperty('firecrawlApiKey', 'test-api-key');
// Optional URL should not be in the result
expect(result.authResult.firecrawlApiUrl).toBeUndefined();
});
it('should preserve safeSearch setting from webSearchConfig', async () => {
// Mock successful authentication
mockLoadAuthValues.mockImplementation(({ authFields }) => {
const result: Record<string, string> = {};
authFields.forEach((field) => {
result[field] = 'test-api-key';
});
return Promise.resolve(result);
});
// Test with safeSearch: false
const configWithSafeSearchOff = { ...webSearchConfig, safeSearch: false } as TWebSearchConfig;
const result = await loadWebSearchAuth({
userId,
webSearchConfig: configWithSafeSearchOff,
loadAuthValues: mockLoadAuthValues,
});
expect(result.authResult).toHaveProperty('safeSearch', false);
});
it('should set the correct service types in authResult', async () => {
// Mock successful authentication
mockLoadAuthValues.mockImplementation(({ authFields }) => {
const result: Record<string, string> = {};
authFields.forEach((field) => {
result[field] =
field === 'FIRECRAWL_API_URL' ? 'https://api.firecrawl.dev' : 'test-api-key';
});
return Promise.resolve(result);
});
const result = await loadWebSearchAuth({
userId,
webSearchConfig,
loadAuthValues: mockLoadAuthValues,
});
// Check that the correct service types are set
expect(result.authResult.searchProvider).toBe('serper' as SearchProviders);
expect(result.authResult.scraperType).toBe('firecrawl' as ScraperTypes);
// One of the rerankers should be set
expect(['jina', 'cohere']).toContain(result.authResult.rerankerType as string);
});
it('should check all services if none are specified', async () => {
// Initialize a webSearchConfig without specific services
const webSearchConfig: TCustomConfig['webSearch'] = {
serperApiKey: '${SERPER_API_KEY}',
firecrawlApiKey: '${FIRECRAWL_API_KEY}',
firecrawlApiUrl: '${FIRECRAWL_API_URL}',
jinaApiKey: '${JINA_API_KEY}',
cohereApiKey: '${COHERE_API_KEY}',
safeSearch: true,
};
// Mock successful authentication
mockLoadAuthValues.mockImplementation(({ authFields }) => {
const result: Record<string, string> = {};
authFields.forEach((field) => {
result[field] =
field === 'FIRECRAWL_API_URL' ? 'https://api.firecrawl.dev' : 'test-api-key';
});
return Promise.resolve(result);
});
const result = await loadWebSearchAuth({
userId,
webSearchConfig,
loadAuthValues: mockLoadAuthValues,
});
expect(result.authenticated).toBe(true);
// Should have checked all categories
expect(result.authTypes).toHaveLength(3);
// Should have set values for all categories
expect(result.authResult.searchProvider).toBeDefined();
expect(result.authResult.scraperType).toBeDefined();
expect(result.authResult.rerankerType).toBeDefined();
});
it('should correctly identify authTypes based on specific configurations', async () => {
// Set up environment variables for system-defined auth
const originalEnv = process.env;
process.env = {
...originalEnv,
SERPER_API_KEY: 'system-serper-key',
FIRECRAWL_API_KEY: 'system-firecrawl-key',
FIRECRAWL_API_URL: 'https://api.firecrawl.dev',
JINA_API_KEY: 'system-jina-key',
COHERE_API_KEY: 'system-cohere-key',
};
// Initialize webSearchConfig with environment variable references
const webSearchConfig: TCustomConfig['webSearch'] = {
serperApiKey: '${SERPER_API_KEY}',
firecrawlApiKey: '${FIRECRAWL_API_KEY}',
firecrawlApiUrl: '${FIRECRAWL_API_URL}',
jinaApiKey: '${JINA_API_KEY}',
cohereApiKey: '${COHERE_API_KEY}',
safeSearch: true,
// Specify which services to use
searchProvider: 'serper' as SearchProviders,
scraperType: 'firecrawl' as ScraperTypes,
rerankerType: 'jina' as RerankerTypes,
};
// Mock loadAuthValues to return the actual values
mockLoadAuthValues.mockImplementation(({ authFields }) => {
const result: Record<string, string> = {};
authFields.forEach((field) => {
if (field === 'SERPER_API_KEY') {
result[field] = 'system-serper-key';
} else if (field === 'FIRECRAWL_API_KEY') {
result[field] = 'system-firecrawl-key';
} else if (field === 'FIRECRAWL_API_URL') {
result[field] = 'https://api.firecrawl.dev';
} else if (field === 'JINA_API_KEY') {
result[field] = 'system-jina-key';
} else if (field === 'COHERE_API_KEY') {
result[field] = 'system-cohere-key';
}
});
return Promise.resolve(result);
});
const result = await loadWebSearchAuth({
userId,
webSearchConfig,
loadAuthValues: mockLoadAuthValues,
});
// Verify that all required fields are present in the authResult
expect(result.authResult).toHaveProperty('serperApiKey');
expect(result.authResult).toHaveProperty('firecrawlApiKey');
expect(result.authResult).toHaveProperty('firecrawlApiUrl');
expect(result.authResult).toHaveProperty('jinaApiKey');
expect(result.authResult).toHaveProperty('searchProvider');
expect(result.authResult).toHaveProperty('scraperType');
expect(result.authResult).toHaveProperty('rerankerType');
expect(result.authenticated).toBe(true);
// Verify authTypes for each category
const providersAuthType = result.authTypes.find(
([category]) => category === 'providers',
)?.[1];
const scrapersAuthType = result.authTypes.find(([category]) => category === 'scrapers')?.[1];
const rerankersAuthType = result.authTypes.find(
([category]) => category === 'rerankers',
)?.[1];
// All should be system-defined since we're using environment variables
expect(providersAuthType).toBe(AuthType.SYSTEM_DEFINED);
expect(scrapersAuthType).toBe(AuthType.SYSTEM_DEFINED);
expect(rerankersAuthType).toBe(AuthType.SYSTEM_DEFINED);
// Verify the authResult contains the correct values
expect(result.authResult).toHaveProperty('serperApiKey', 'system-serper-key');
expect(result.authResult).toHaveProperty('firecrawlApiKey', 'system-firecrawl-key');
expect(result.authResult).toHaveProperty('firecrawlApiUrl', 'https://api.firecrawl.dev');
expect(result.authResult).toHaveProperty('jinaApiKey', 'system-jina-key');
expect(result.authResult).toHaveProperty('searchProvider', 'serper');
expect(result.authResult).toHaveProperty('scraperType', 'firecrawl');
expect(result.authResult).toHaveProperty('rerankerType', 'jina');
// Restore original env
process.env = originalEnv;
});
it('should handle custom variable names in environment variables', async () => {
// Set up environment variables with custom names
const originalEnv = process.env;
process.env = {
...originalEnv,
CUSTOM_SERPER_KEY: 'custom-serper-key',
CUSTOM_FIRECRAWL_KEY: 'custom-firecrawl-key',
CUSTOM_FIRECRAWL_URL: 'https://custom.firecrawl.dev',
CUSTOM_JINA_KEY: 'custom-jina-key',
CUSTOM_COHERE_KEY: 'custom-cohere-key',
};
// Initialize webSearchConfig with custom variable names
const webSearchConfig: TCustomConfig['webSearch'] = {
serperApiKey: '${CUSTOM_SERPER_KEY}',
firecrawlApiKey: '${CUSTOM_FIRECRAWL_KEY}',
firecrawlApiUrl: '${CUSTOM_FIRECRAWL_URL}',
jinaApiKey: '${CUSTOM_JINA_KEY}',
cohereApiKey: '${CUSTOM_COHERE_KEY}',
safeSearch: true,
// Specify which services to use
searchProvider: 'serper' as SearchProviders,
scraperType: 'firecrawl' as ScraperTypes,
rerankerType: 'jina' as RerankerTypes, // Only Jina will be checked
};
// Mock loadAuthValues to return the actual values
mockLoadAuthValues.mockImplementation(({ authFields }) => {
const result: Record<string, string> = {};
authFields.forEach((field) => {
if (field === 'CUSTOM_SERPER_KEY') {
result[field] = 'custom-serper-key';
} else if (field === 'CUSTOM_FIRECRAWL_KEY') {
result[field] = 'custom-firecrawl-key';
} else if (field === 'CUSTOM_FIRECRAWL_URL') {
result[field] = 'https://custom.firecrawl.dev';
} else if (field === 'CUSTOM_JINA_KEY') {
result[field] = 'custom-jina-key';
}
// Note: CUSTOM_COHERE_KEY is not checked because we specified jina as rerankerType
});
return Promise.resolve(result);
});
const result = await loadWebSearchAuth({
userId,
webSearchConfig,
loadAuthValues: mockLoadAuthValues,
});
expect(result.authenticated).toBe(true);
// Verify the authResult contains the correct values from custom variables
expect(result.authResult).toHaveProperty('serperApiKey', 'custom-serper-key');
expect(result.authResult).toHaveProperty('firecrawlApiKey', 'custom-firecrawl-key');
expect(result.authResult).toHaveProperty('firecrawlApiUrl', 'https://custom.firecrawl.dev');
expect(result.authResult).toHaveProperty('jinaApiKey', 'custom-jina-key');
// cohereApiKey should not be in the result since we specified jina as rerankerType
expect(result.authResult).not.toHaveProperty('cohereApiKey');
// Verify the service types are set correctly
expect(result.authResult).toHaveProperty('searchProvider', 'serper');
expect(result.authResult).toHaveProperty('scraperType', 'firecrawl');
expect(result.authResult).toHaveProperty('rerankerType', 'jina');
// Restore original env
process.env = originalEnv;
});
it('should always return authTypes array with exactly 3 categories', async () => {
// Set up environment variables
const originalEnv = process.env;
process.env = {
...originalEnv,
SERPER_API_KEY: 'test-key',
FIRECRAWL_API_KEY: 'test-key',
FIRECRAWL_API_URL: 'https://api.firecrawl.dev',
JINA_API_KEY: 'test-key',
};
// Initialize webSearchConfig with environment variable references
const webSearchConfig: TCustomConfig['webSearch'] = {
serperApiKey: '${SERPER_API_KEY}',
firecrawlApiKey: '${FIRECRAWL_API_KEY}',
firecrawlApiUrl: '${FIRECRAWL_API_URL}',
jinaApiKey: '${JINA_API_KEY}',
cohereApiKey: '${COHERE_API_KEY}',
safeSearch: true,
};
// Mock loadAuthValues to return values
mockLoadAuthValues.mockImplementation(({ authFields }) => {
const result: Record<string, string> = {};
authFields.forEach((field) => {
result[field] = field === 'FIRECRAWL_API_URL' ? 'https://api.firecrawl.dev' : 'test-key';
});
return Promise.resolve(result);
});
const result = await loadWebSearchAuth({
userId,
webSearchConfig,
loadAuthValues: mockLoadAuthValues,
});
// Get the number of categories from webSearchAuth
const expectedCategoryCount = Object.keys(webSearchAuth).length;
// Verify authTypes array structure
expect(result.authTypes).toHaveLength(expectedCategoryCount);
// Verify each category exists exactly once
const categories = result.authTypes.map(([category]) => category);
Object.keys(webSearchAuth).forEach((category) => {
expect(categories).toContain(category);
});
// Verify no duplicate categories
expect(new Set(categories).size).toBe(expectedCategoryCount);
// Verify each entry has the correct format [category, AuthType]
result.authTypes.forEach(([category, authType]) => {
expect(typeof category).toBe('string');
expect([AuthType.SYSTEM_DEFINED, AuthType.USER_PROVIDED]).toContain(authType);
});
// Restore original env
process.env = originalEnv;
});
it('should maintain authTypes array structure even when authentication fails', async () => {
// Set up environment variables
const originalEnv = process.env;
process.env = {
...originalEnv,
SERPER_API_KEY: 'test-key',
// Missing other keys to force authentication failure
};
// Initialize webSearchConfig with environment variable references
const webSearchConfig: TCustomConfig['webSearch'] = {
serperApiKey: '${SERPER_API_KEY}',
firecrawlApiKey: '${FIRECRAWL_API_KEY}',
firecrawlApiUrl: '${FIRECRAWL_API_URL}',
jinaApiKey: '${JINA_API_KEY}',
cohereApiKey: '${COHERE_API_KEY}',
safeSearch: true,
};
// Mock loadAuthValues to return partial values
mockLoadAuthValues.mockImplementation(({ authFields }) => {
const result: Record<string, string> = {};
authFields.forEach((field) => {
if (field === 'SERPER_API_KEY') {
result[field] = 'test-key';
}
// Other fields are intentionally missing
});
return Promise.resolve(result);
});
const result = await loadWebSearchAuth({
userId,
webSearchConfig,
loadAuthValues: mockLoadAuthValues,
});
// Get the number of categories from webSearchAuth
const expectedCategoryCount = Object.keys(webSearchAuth).length;
// Verify authentication failed
expect(result.authenticated).toBe(false);
// Verify authTypes array structure is maintained
expect(result.authTypes).toHaveLength(expectedCategoryCount);
// Verify each category exists exactly once
const categories = result.authTypes.map(([category]) => category);
Object.keys(webSearchAuth).forEach((category) => {
expect(categories).toContain(category);
});
// Verify no duplicate categories
expect(new Set(categories).size).toBe(expectedCategoryCount);
// Verify each entry has the correct format [category, AuthType]
result.authTypes.forEach(([category, authType]) => {
expect(typeof category).toBe('string');
expect([AuthType.SYSTEM_DEFINED, AuthType.USER_PROVIDED]).toContain(authType);
});
// Restore original env
process.env = originalEnv;
});
});
describe('webSearchAuth', () => {
it('should have the expected structure', () => {
// Check that all expected categories exist
expect(webSearchAuth).toHaveProperty('providers');
expect(webSearchAuth).toHaveProperty('scrapers');
expect(webSearchAuth).toHaveProperty('rerankers');
// Check providers
expect(webSearchAuth.providers).toHaveProperty('serper');
expect(webSearchAuth.providers.serper).toHaveProperty('serperApiKey', 1);
// Check scrapers
expect(webSearchAuth.scrapers).toHaveProperty('firecrawl');
expect(webSearchAuth.scrapers.firecrawl).toHaveProperty('firecrawlApiKey', 1);
expect(webSearchAuth.scrapers.firecrawl).toHaveProperty('firecrawlApiUrl', 0);
// Check rerankers
expect(webSearchAuth.rerankers).toHaveProperty('jina');
expect(webSearchAuth.rerankers.jina).toHaveProperty('jinaApiKey', 1);
expect(webSearchAuth.rerankers).toHaveProperty('cohere');
expect(webSearchAuth.rerankers.cohere).toHaveProperty('cohereApiKey', 1);
});
it('should mark required keys with value 1', () => {
// All keys with value 1 are required
expect(webSearchAuth.providers.serper.serperApiKey).toBe(1);
expect(webSearchAuth.scrapers.firecrawl.firecrawlApiKey).toBe(1);
expect(webSearchAuth.rerankers.jina.jinaApiKey).toBe(1);
expect(webSearchAuth.rerankers.cohere.cohereApiKey).toBe(1);
});
it('should mark optional keys with value 0', () => {
// Keys with value 0 are optional
expect(webSearchAuth.scrapers.firecrawl.firecrawlApiUrl).toBe(0);
});
});
describe('loadWebSearchAuth with specific services', () => {
// Common test variables
const userId = 'test-user-id';
let mockLoadAuthValues: jest.Mock;
beforeEach(() => {
// Reset mocks before each test
jest.clearAllMocks();
// Initialize the mock function
mockLoadAuthValues = jest.fn();
});
it('should only check the specified searchProvider', async () => {
// Initialize a webSearchConfig with a specific searchProvider
const webSearchConfig: TCustomConfig['webSearch'] = {
serperApiKey: '${SERPER_API_KEY}',
firecrawlApiKey: '${FIRECRAWL_API_KEY}',
firecrawlApiUrl: '${FIRECRAWL_API_URL}',
jinaApiKey: '${JINA_API_KEY}',
cohereApiKey: '${COHERE_API_KEY}',
safeSearch: true,
searchProvider: 'serper' as SearchProviders,
};
// Mock successful authentication
mockLoadAuthValues.mockImplementation(({ authFields }) => {
const result: Record<string, string> = {};
authFields.forEach((field) => {
result[field] =
field === 'FIRECRAWL_API_URL' ? 'https://api.firecrawl.dev' : 'test-api-key';
});
return Promise.resolve(result);
});
const result = await loadWebSearchAuth({
userId,
webSearchConfig,
loadAuthValues: mockLoadAuthValues,
});
expect(result.authenticated).toBe(true);
expect(result.authResult.searchProvider).toBe('serper');
// Verify that only SERPER_API_KEY was requested for the providers category
const providerCalls = mockLoadAuthValues.mock.calls.filter((call) =>
call[0].authFields.includes('SERPER_API_KEY'),
);
expect(providerCalls.length).toBe(1);
});
it('should only check the specified scraperType', async () => {
// Initialize a webSearchConfig with a specific scraperType
const webSearchConfig: TCustomConfig['webSearch'] = {
serperApiKey: '${SERPER_API_KEY}',
firecrawlApiKey: '${FIRECRAWL_API_KEY}',
firecrawlApiUrl: '${FIRECRAWL_API_URL}',
jinaApiKey: '${JINA_API_KEY}',
cohereApiKey: '${COHERE_API_KEY}',
safeSearch: true,
scraperType: 'firecrawl' as ScraperTypes,
};
// Mock successful authentication
mockLoadAuthValues.mockImplementation(({ authFields }) => {
const result: Record<string, string> = {};
authFields.forEach((field) => {
result[field] =
field === 'FIRECRAWL_API_URL' ? 'https://api.firecrawl.dev' : 'test-api-key';
});
return Promise.resolve(result);
});
const result = await loadWebSearchAuth({
userId,
webSearchConfig,
loadAuthValues: mockLoadAuthValues,
});
expect(result.authenticated).toBe(true);
expect(result.authResult.scraperType).toBe('firecrawl');
// Verify that only FIRECRAWL_API_KEY and FIRECRAWL_API_URL were requested for the scrapers category
const scraperCalls = mockLoadAuthValues.mock.calls.filter((call) =>
call[0].authFields.includes('FIRECRAWL_API_KEY'),
);
expect(scraperCalls.length).toBe(1);
});
it('should only check the specified rerankerType', async () => {
// Initialize a webSearchConfig with a specific rerankerType
const webSearchConfig: TCustomConfig['webSearch'] = {
serperApiKey: '${SERPER_API_KEY}',
firecrawlApiKey: '${FIRECRAWL_API_KEY}',
firecrawlApiUrl: '${FIRECRAWL_API_URL}',
jinaApiKey: '${JINA_API_KEY}',
cohereApiKey: '${COHERE_API_KEY}',
safeSearch: true,
rerankerType: 'jina' as RerankerTypes,
};
// Mock successful authentication
mockLoadAuthValues.mockImplementation(({ authFields }) => {
const result: Record<string, string> = {};
authFields.forEach((field) => {
result[field] =
field === 'FIRECRAWL_API_URL' ? 'https://api.firecrawl.dev' : 'test-api-key';
});
return Promise.resolve(result);
});
const result = await loadWebSearchAuth({
userId,
webSearchConfig,
loadAuthValues: mockLoadAuthValues,
});
expect(result.authenticated).toBe(true);
expect(result.authResult.rerankerType).toBe('jina');
// Verify that only JINA_API_KEY was requested for the rerankers category
const rerankerCalls = mockLoadAuthValues.mock.calls.filter((call) =>
call[0].authFields.includes('JINA_API_KEY'),
);
expect(rerankerCalls.length).toBe(1);
// Verify that COHERE_API_KEY was not requested
const cohereCalls = mockLoadAuthValues.mock.calls.filter((call) =>
call[0].authFields.includes('COHERE_API_KEY'),
);
expect(cohereCalls.length).toBe(0);
});
it('should handle invalid specified service gracefully', async () => {
// Initialize a webSearchConfig with an invalid searchProvider
const webSearchConfig: TCustomConfig['webSearch'] = {
serperApiKey: '${SERPER_API_KEY}',
firecrawlApiKey: '${FIRECRAWL_API_KEY}',
firecrawlApiUrl: '${FIRECRAWL_API_URL}',
jinaApiKey: '${JINA_API_KEY}',
cohereApiKey: '${COHERE_API_KEY}',
safeSearch: true,
searchProvider: 'invalid-provider' as SearchProviders,
};
// Mock successful authentication
mockLoadAuthValues.mockImplementation(({ authFields }) => {
const result: Record<string, string> = {};
authFields.forEach((field) => {
result[field] =
field === 'FIRECRAWL_API_URL' ? 'https://api.firecrawl.dev' : 'test-api-key';
});
return Promise.resolve(result);
});
const result = await loadWebSearchAuth({
userId,
webSearchConfig,
loadAuthValues: mockLoadAuthValues,
});
// Should fail because the specified provider doesn't exist
expect(result.authenticated).toBe(false);
});
it('should fail authentication when specified service is not authenticated but others are', async () => {
// Initialize a webSearchConfig with a specific rerankerType (jina)
const webSearchConfig: TCustomConfig['webSearch'] = {
serperApiKey: '${SERPER_API_KEY}',
firecrawlApiKey: '${FIRECRAWL_API_KEY}',
firecrawlApiUrl: '${FIRECRAWL_API_URL}',
jinaApiKey: '${JINA_API_KEY}',
cohereApiKey: '${COHERE_API_KEY}',
safeSearch: true,
rerankerType: 'jina' as RerankerTypes,
};
// Mock authentication where cohere is authenticated but jina is not
mockLoadAuthValues.mockImplementation(({ authFields }) => {
const result: Record<string, string> = {};
authFields.forEach((field) => {
// Authenticate all fields except JINA_API_KEY
if (field !== 'JINA_API_KEY') {
result[field] =
field === 'FIRECRAWL_API_URL' ? 'https://api.firecrawl.dev' : 'test-api-key';
}
});
return Promise.resolve(result);
});
const result = await loadWebSearchAuth({
userId,
webSearchConfig,
loadAuthValues: mockLoadAuthValues,
});
// Should fail because the specified reranker (jina) is not authenticated
// even though another reranker (cohere) might be authenticated
expect(result.authenticated).toBe(false);
// Verify that JINA_API_KEY was requested
const jinaApiKeyCalls = mockLoadAuthValues.mock.calls.filter((call) =>
call[0].authFields.includes('JINA_API_KEY'),
);
expect(jinaApiKeyCalls.length).toBe(1);
// Verify that COHERE_API_KEY was not requested since we specified jina
const cohereApiKeyCalls = mockLoadAuthValues.mock.calls.filter((call) =>
call[0].authFields.includes('COHERE_API_KEY'),
);
expect(cohereApiKeyCalls.length).toBe(0);
});
it('should check all services if none are specified', async () => {
// Initialize a webSearchConfig without specific services
const webSearchConfig: TCustomConfig['webSearch'] = {
serperApiKey: '${SERPER_API_KEY}',
firecrawlApiKey: '${FIRECRAWL_API_KEY}',
firecrawlApiUrl: '${FIRECRAWL_API_URL}',
jinaApiKey: '${JINA_API_KEY}',
cohereApiKey: '${COHERE_API_KEY}',
safeSearch: true,
};
// Mock successful authentication
mockLoadAuthValues.mockImplementation(({ authFields }) => {
const result: Record<string, string> = {};
authFields.forEach((field) => {
result[field] =
field === 'FIRECRAWL_API_URL' ? 'https://api.firecrawl.dev' : 'test-api-key';
});
return Promise.resolve(result);
});
const result = await loadWebSearchAuth({
userId,
webSearchConfig,
loadAuthValues: mockLoadAuthValues,
});
expect(result.authenticated).toBe(true);
// Should have checked all categories
expect(result.authTypes).toHaveLength(3);
// Should have set values for all categories
expect(result.authResult.searchProvider).toBeDefined();
expect(result.authResult.scraperType).toBeDefined();
expect(result.authResult.rerankerType).toBeDefined();
});
});
});