@retrivora-ai/rag-engine 2.3.6 → 2.3.7

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -1,5 +1,5 @@
1
1
  import { CSSProperties, MouseEvent, ReactNode, ElementType } from 'react';
2
- import { W as WidgetPosition, Z as Product, s as UIConfig, P as ProjectChatSettings, o as RagMessage, w as VectorMatch, J as UITransformationResponse, n as RagConfig, I as ILLMProvider, f as IngestDocument, C as ChatMessage, c as ChatResponse, _ as RetrievalResult, t as UniversalRagConfig } from './ILLMProvider-B8TAVx3j.mjs';
2
+ import { W as WidgetPosition, Z as Product, s as UIConfig, P as ProjectChatSettings, o as RagMessage, w as VectorMatch, L as LLMConfig, d as EmbeddingConfig, V as VectorDBConfig, J as UITransformationResponse, n as RagConfig, I as ILLMProvider, f as IngestDocument, C as ChatMessage, c as ChatResponse, _ as RetrievalResult, t as UniversalRagConfig } from './ILLMProvider-B8TAVx3j.mjs';
3
3
 
4
4
  interface ArchitectureCardProps {
5
5
  icon: ReactNode;
@@ -219,6 +219,104 @@ declare class ConfigFetcher {
219
219
  static fetchRemoteVectorConfig(projectId: string, licenseKey?: string): Promise<RemoteVectorConfig | null>;
220
220
  }
221
221
 
222
+ /**
223
+ * LicenseConfigResolver — Resolves LLM and embedding configuration
224
+ * based on the user's license tier from Retrivora admin dashboard.
225
+ *
226
+ * The end-user SDK only needs:
227
+ * - licenseKey (JWT token from Retrivora)
228
+ * - projectId (workspace identifier)
229
+ *
230
+ * Configuration includes:
231
+ * 1. Tier-based model selection (/api/v1/config)
232
+ * 2. Provider API credentials (/api/v1/credentials)
233
+ * 3. Vector DB and rate limit settings
234
+ *
235
+ * All provider API keys (Gemini, Groq, Pinecone) are stored on Retrivora backend
236
+ * and never exposed to end-users directly.
237
+ */
238
+
239
+ interface LicenseTierConfig {
240
+ tier: 'free_trial' | 'free' | 'hobby' | 'pro' | 'enterprise';
241
+ llmModel: string;
242
+ embeddingModel: string;
243
+ embeddingDimensions: number;
244
+ maxTokens: number;
245
+ chunkSize: number;
246
+ chunkOverlap: number;
247
+ topK: number;
248
+ vectorDb: {
249
+ provider: string;
250
+ indexName: string;
251
+ };
252
+ rateLimit?: {
253
+ requestsPerMinute: number;
254
+ tokensPerDay: number;
255
+ };
256
+ }
257
+ interface ProviderCredentials {
258
+ geminiApiKey?: string;
259
+ groqApiKey?: string;
260
+ pineconeApiKey?: string;
261
+ textEmbeddingProvider?: string;
262
+ }
263
+ interface ProviderCredentials {
264
+ geminiApiKey?: string;
265
+ groqApiKey?: string;
266
+ pineconeApiKey?: string;
267
+ textEmbeddingProvider?: string;
268
+ }
269
+ /**
270
+ * Resolver for license-based configuration.
271
+ * This is called by the end-user SDK after they instantiate with licenseKey + projectId.
272
+ */
273
+ declare class LicenseConfigResolver {
274
+ private static readonly RETRIVORA_CONFIG_URL;
275
+ private static readonly RETRIVORA_CREDENTIALS_URL;
276
+ private static readonly CACHE_TTL_MS;
277
+ private static configCache;
278
+ /**
279
+ * Validate and resolve configuration from license key
280
+ * @param licenseKey JWT token from Retrivora (format: rtv_eyJ...)
281
+ * @param projectId Workspace identifier
282
+ * @returns Resolved LLM and embedding configs with provider credentials
283
+ */
284
+ static resolveFromLicense(licenseKey: string, projectId: string): Promise<{
285
+ llmConfig: LLMConfig;
286
+ embeddingConfig: EmbeddingConfig;
287
+ vectorDbConfig: VectorDBConfig;
288
+ tierConfig: LicenseTierConfig;
289
+ credentials: ProviderCredentials;
290
+ }>;
291
+ /**
292
+ * Fetch tier configuration from Retrivora config API
293
+ */
294
+ private static fetchTierConfig;
295
+ /**
296
+ * Fetch provider credentials from Retrivora credentials API
297
+ */
298
+ private static fetchProviderCredentials;
299
+ /**
300
+ * Build LLM, embedding, and vector DB configs from tier and credentials
301
+ */
302
+ private static buildConfigs;
303
+ /**
304
+ * Get provider-specific vector DB options
305
+ */
306
+ private static getVectorDbOptions;
307
+ /**
308
+ * Clear configuration cache (useful for testing or manual refresh)
309
+ */
310
+ static clearCache(): void;
311
+ /**
312
+ * Get cache stats for debugging
313
+ */
314
+ static getCacheStats(): {
315
+ size: number;
316
+ entries: string[];
317
+ };
318
+ }
319
+
222
320
  /**
223
321
  * Primary rendering outcome types supported by the decision engine.
224
322
  */
@@ -661,4 +759,4 @@ declare class BatchProcessor {
661
759
  static processConcurrent<T, R>(items: T[], processor: (item: T) => Promise<R>, concurrency?: number, options?: BatchOptions): Promise<BatchResult<R>>;
662
760
  }
663
761
 
664
- export { AuthenticationException as A, type BatchOptions as B, type ChatWidgetProps as C, type DocumentUploadProps as D, EmbeddingFailedException as E, RetrivoraError as F, type RetrivoraErrorCode as G, SDKVersionUnsupportedError as H, type IRenderRule as I, SDK_VERSION as J, isTransientError as K, LicenseValidationError as L, type MessageBubbleProps as M, type ArchitectureCardProps as N, type ChatViewportSize as O, type ProductCardProps as P, DocumentChunker as Q, type RemoteConfig as R, type SourceCardProps as S, Pipeline as T, type PipelineStep as U, type ProviderPill as V, type RenderSectionDecision as W, type Snippet as X, wrapError as Y, type ChatWindowProps as a, type ConfigProviderProps as b, type ClientConfig as c, type ProductCarouselProps as d, BatchProcessor as e, type BatchResult as f, type ChartType as g, type Chunk as h, type ChunkOptions as i, ConfigFetcher as j, ConfigurationException as k, type DecisionContext as l, type IRendererStrategy as m, type IntentCategory as n, type LicenseValidationRequest as o, type LicenseValidationResponse as p, LicenseValidator as q, ProviderNotFoundException as r, RateLimitException as s, type RemoteEmbeddingConfig as t, type RemoteLLMConfig as u, type RemoteVectorConfig as v, type RenderDecision as w, type RenderType as x, RetrievalException as y, Retrivora as z };
762
+ export { wrapError as $, AuthenticationException as A, type BatchOptions as B, type ChatWidgetProps as C, type DocumentUploadProps as D, EmbeddingFailedException as E, RetrievalException as F, Retrivora as G, RetrivoraError as H, type IRenderRule as I, type RetrivoraErrorCode as J, SDKVersionUnsupportedError as K, type LicenseTierConfig as L, type MessageBubbleProps as M, SDK_VERSION as N, isTransientError as O, type ProductCardProps as P, type ArchitectureCardProps as Q, type RemoteConfig as R, type SourceCardProps as S, type ChatViewportSize as T, DocumentChunker as U, LicenseConfigResolver as V, Pipeline as W, type PipelineStep as X, type ProviderPill as Y, type RenderSectionDecision as Z, type Snippet as _, type ChatWindowProps as a, type ConfigProviderProps as b, type ClientConfig as c, type ProductCarouselProps as d, BatchProcessor as e, type BatchResult as f, type ChartType as g, type Chunk as h, type ChunkOptions as i, ConfigFetcher as j, ConfigurationException as k, type DecisionContext as l, type IRendererStrategy as m, type IntentCategory as n, LicenseValidationError as o, type LicenseValidationRequest as p, type LicenseValidationResponse as q, LicenseValidator as r, type ProviderCredentials as s, ProviderNotFoundException as t, RateLimitException as u, type RemoteEmbeddingConfig as v, type RemoteLLMConfig as w, type RemoteVectorConfig as x, type RenderDecision as y, type RenderType as z };
@@ -1,5 +1,5 @@
1
1
  import { CSSProperties, MouseEvent, ReactNode, ElementType } from 'react';
2
- import { W as WidgetPosition, Z as Product, s as UIConfig, P as ProjectChatSettings, o as RagMessage, w as VectorMatch, J as UITransformationResponse, n as RagConfig, I as ILLMProvider, f as IngestDocument, C as ChatMessage, c as ChatResponse, _ as RetrievalResult, t as UniversalRagConfig } from './ILLMProvider-B8TAVx3j.js';
2
+ import { W as WidgetPosition, Z as Product, s as UIConfig, P as ProjectChatSettings, o as RagMessage, w as VectorMatch, L as LLMConfig, d as EmbeddingConfig, V as VectorDBConfig, J as UITransformationResponse, n as RagConfig, I as ILLMProvider, f as IngestDocument, C as ChatMessage, c as ChatResponse, _ as RetrievalResult, t as UniversalRagConfig } from './ILLMProvider-B8TAVx3j.js';
3
3
 
4
4
  interface ArchitectureCardProps {
5
5
  icon: ReactNode;
@@ -219,6 +219,104 @@ declare class ConfigFetcher {
219
219
  static fetchRemoteVectorConfig(projectId: string, licenseKey?: string): Promise<RemoteVectorConfig | null>;
220
220
  }
221
221
 
222
+ /**
223
+ * LicenseConfigResolver — Resolves LLM and embedding configuration
224
+ * based on the user's license tier from Retrivora admin dashboard.
225
+ *
226
+ * The end-user SDK only needs:
227
+ * - licenseKey (JWT token from Retrivora)
228
+ * - projectId (workspace identifier)
229
+ *
230
+ * Configuration includes:
231
+ * 1. Tier-based model selection (/api/v1/config)
232
+ * 2. Provider API credentials (/api/v1/credentials)
233
+ * 3. Vector DB and rate limit settings
234
+ *
235
+ * All provider API keys (Gemini, Groq, Pinecone) are stored on Retrivora backend
236
+ * and never exposed to end-users directly.
237
+ */
238
+
239
+ interface LicenseTierConfig {
240
+ tier: 'free_trial' | 'free' | 'hobby' | 'pro' | 'enterprise';
241
+ llmModel: string;
242
+ embeddingModel: string;
243
+ embeddingDimensions: number;
244
+ maxTokens: number;
245
+ chunkSize: number;
246
+ chunkOverlap: number;
247
+ topK: number;
248
+ vectorDb: {
249
+ provider: string;
250
+ indexName: string;
251
+ };
252
+ rateLimit?: {
253
+ requestsPerMinute: number;
254
+ tokensPerDay: number;
255
+ };
256
+ }
257
+ interface ProviderCredentials {
258
+ geminiApiKey?: string;
259
+ groqApiKey?: string;
260
+ pineconeApiKey?: string;
261
+ textEmbeddingProvider?: string;
262
+ }
263
+ interface ProviderCredentials {
264
+ geminiApiKey?: string;
265
+ groqApiKey?: string;
266
+ pineconeApiKey?: string;
267
+ textEmbeddingProvider?: string;
268
+ }
269
+ /**
270
+ * Resolver for license-based configuration.
271
+ * This is called by the end-user SDK after they instantiate with licenseKey + projectId.
272
+ */
273
+ declare class LicenseConfigResolver {
274
+ private static readonly RETRIVORA_CONFIG_URL;
275
+ private static readonly RETRIVORA_CREDENTIALS_URL;
276
+ private static readonly CACHE_TTL_MS;
277
+ private static configCache;
278
+ /**
279
+ * Validate and resolve configuration from license key
280
+ * @param licenseKey JWT token from Retrivora (format: rtv_eyJ...)
281
+ * @param projectId Workspace identifier
282
+ * @returns Resolved LLM and embedding configs with provider credentials
283
+ */
284
+ static resolveFromLicense(licenseKey: string, projectId: string): Promise<{
285
+ llmConfig: LLMConfig;
286
+ embeddingConfig: EmbeddingConfig;
287
+ vectorDbConfig: VectorDBConfig;
288
+ tierConfig: LicenseTierConfig;
289
+ credentials: ProviderCredentials;
290
+ }>;
291
+ /**
292
+ * Fetch tier configuration from Retrivora config API
293
+ */
294
+ private static fetchTierConfig;
295
+ /**
296
+ * Fetch provider credentials from Retrivora credentials API
297
+ */
298
+ private static fetchProviderCredentials;
299
+ /**
300
+ * Build LLM, embedding, and vector DB configs from tier and credentials
301
+ */
302
+ private static buildConfigs;
303
+ /**
304
+ * Get provider-specific vector DB options
305
+ */
306
+ private static getVectorDbOptions;
307
+ /**
308
+ * Clear configuration cache (useful for testing or manual refresh)
309
+ */
310
+ static clearCache(): void;
311
+ /**
312
+ * Get cache stats for debugging
313
+ */
314
+ static getCacheStats(): {
315
+ size: number;
316
+ entries: string[];
317
+ };
318
+ }
319
+
222
320
  /**
223
321
  * Primary rendering outcome types supported by the decision engine.
224
322
  */
@@ -661,4 +759,4 @@ declare class BatchProcessor {
661
759
  static processConcurrent<T, R>(items: T[], processor: (item: T) => Promise<R>, concurrency?: number, options?: BatchOptions): Promise<BatchResult<R>>;
662
760
  }
663
761
 
664
- export { AuthenticationException as A, type BatchOptions as B, type ChatWidgetProps as C, type DocumentUploadProps as D, EmbeddingFailedException as E, RetrivoraError as F, type RetrivoraErrorCode as G, SDKVersionUnsupportedError as H, type IRenderRule as I, SDK_VERSION as J, isTransientError as K, LicenseValidationError as L, type MessageBubbleProps as M, type ArchitectureCardProps as N, type ChatViewportSize as O, type ProductCardProps as P, DocumentChunker as Q, type RemoteConfig as R, type SourceCardProps as S, Pipeline as T, type PipelineStep as U, type ProviderPill as V, type RenderSectionDecision as W, type Snippet as X, wrapError as Y, type ChatWindowProps as a, type ConfigProviderProps as b, type ClientConfig as c, type ProductCarouselProps as d, BatchProcessor as e, type BatchResult as f, type ChartType as g, type Chunk as h, type ChunkOptions as i, ConfigFetcher as j, ConfigurationException as k, type DecisionContext as l, type IRendererStrategy as m, type IntentCategory as n, type LicenseValidationRequest as o, type LicenseValidationResponse as p, LicenseValidator as q, ProviderNotFoundException as r, RateLimitException as s, type RemoteEmbeddingConfig as t, type RemoteLLMConfig as u, type RemoteVectorConfig as v, type RenderDecision as w, type RenderType as x, RetrievalException as y, Retrivora as z };
762
+ export { wrapError as $, AuthenticationException as A, type BatchOptions as B, type ChatWidgetProps as C, type DocumentUploadProps as D, EmbeddingFailedException as E, RetrievalException as F, Retrivora as G, RetrivoraError as H, type IRenderRule as I, type RetrivoraErrorCode as J, SDKVersionUnsupportedError as K, type LicenseTierConfig as L, type MessageBubbleProps as M, SDK_VERSION as N, isTransientError as O, type ProductCardProps as P, type ArchitectureCardProps as Q, type RemoteConfig as R, type SourceCardProps as S, type ChatViewportSize as T, DocumentChunker as U, LicenseConfigResolver as V, Pipeline as W, type PipelineStep as X, type ProviderPill as Y, type RenderSectionDecision as Z, type Snippet as _, type ChatWindowProps as a, type ConfigProviderProps as b, type ClientConfig as c, type ProductCarouselProps as d, BatchProcessor as e, type BatchResult as f, type ChartType as g, type Chunk as h, type ChunkOptions as i, ConfigFetcher as j, ConfigurationException as k, type DecisionContext as l, type IRendererStrategy as m, type IntentCategory as n, LicenseValidationError as o, type LicenseValidationRequest as p, type LicenseValidationResponse as q, LicenseValidator as r, type ProviderCredentials as s, ProviderNotFoundException as t, RateLimitException as u, type RemoteEmbeddingConfig as v, type RemoteLLMConfig as w, type RemoteVectorConfig as x, type RenderDecision as y, type RenderType as z };
@@ -317,7 +317,7 @@ var init_package = __esm({
317
317
  "package.json"() {
318
318
  package_default = {
319
319
  name: "@retrivora-ai/rag-engine",
320
- version: "2.3.6",
320
+ version: "2.3.7",
321
321
  description: "Retrivora AI is a plug-and-play AI engine for RAG chat experiences \u2014 generic vector DB + LLM provider, embeddable or standalone.",
322
322
  author: "Abhinav Alkuchi",
323
323
  license: "UNLICENSED",
@@ -2849,13 +2849,13 @@ LicenseVerifier.CACHE_TTL_MS = 60 * 1e3;
2849
2849
  // In production, this matches the private key held by Retrivora SaaS.
2850
2850
  // Rotate this key only when the signing private key is replaced.
2851
2851
  LicenseVerifier.PUBLIC_KEY = `-----BEGIN PUBLIC KEY-----
2852
- MIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEAoOQlmDSKfCZaRy06rxwa
2853
- chuDimlF134xTCitKDTPTPizF23oSn34pGDJakm/0T09JVUX8L/cvhGPe/3lf/g9
2854
- 4Y1q+Gb2KVM17KNGeT1Aj9Ua3BKz/yao/oFhi/r3Zt3cWY13B/Rd6clqY9WshIAV
2855
- NqZNM8ilTl777cMbyym+6D+mzGXgmnJAO5brddIAI9FPLW0T3Znt8vEqDOIljEkC
2856
- HCGHeirKABLbjKSJ4hljCa3vVitWqndhiq8CBSJG6HatiSGYotgMWGRY+0EbdlEv
2857
- OJP4SqafQ3aD6YnL9q1n15QKwxu/uMK+JGM2VxxuyUi9GANtMIkCYy9xya73P4Qc
2858
- gQIDAQAB
2852
+ MIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEAyKvvOSYdCimG3b236ebV
2853
+ jpG88xQiU8dSiaMahtBJDf4w0AewRj7hMNgmeZ1wyTOj8kVgeGBOolp1Z0QmTr6O
2854
+ Tj4EDQ/zjsg12If7hSf3NiwLHCYmMLslyK0aaP6PQoTrYkBv3GNdkMecrIjKCijb
2855
+ 6VvDLLF3X696melTVM0z/d2e50BW2AlRw/rf1EqYi10d0BcPo5zje0XETr/WGmqu
2856
+ sDiXtsP5MlwkL9X7RQY2uovRa+l/vOAqAfdmJ+OzXuk5P7mDUPb+gCtKnoDbb1tr
2857
+ 6UAu733dkUnetAcMqyT3fNPBLO66DCc6m9rGCYYAaG7piMEDwG0/ZySZBFbgllJc
2858
+ 6QIDAQAB
2859
2859
  -----END PUBLIC KEY-----`;
2860
2860
 
2861
2861
  // src/config/constants.ts
@@ -2904,6 +2904,7 @@ var PROVIDERS_WITH_EMBEDDINGS = [
2904
2904
  var UI_BORDER_RADIUS_OPTIONS = ["none", "sm", "md", "lg", "xl", "full"];
2905
2905
 
2906
2906
  // src/config/serverConfig.ts
2907
+ var _runtimeConfig = null;
2907
2908
  function readString(env, name) {
2908
2909
  var _a3;
2909
2910
  const value = (_a3 = env[name]) == null ? void 0 : _a3.trim();
@@ -2927,7 +2928,7 @@ function readEnum(env, name, fallback, allowed) {
2927
2928
  throw new Error(`[getRagConfig] ${name} must be one of: ${allowed.join(", ")}`);
2928
2929
  }
2929
2930
  function getEnvConfig(env = process.env, base) {
2930
- var _a3, _b, _c, _d, _e, _f, _g2, _h, _i, _j, _k, _l, _m, _n, _o, _p, _q, _r, _s, _t, _u, _v, _w, _x, _y, _z, _A, _B, _C, _D, _E, _F, _G, _H, _I, _J, _K, _L, _M, _N, _O, _P, _Q, _R, _S, _T, _U, _V, _W, _X, _Y, _Z, __, _$, _aa, _ba, _ca, _da, _ea, _fa, _ga, _ha, _ia, _ja, _ka, _la, _ma, _na, _oa, _pa, _qa, _ra, _sa, _ta, _ua, _va, _wa, _xa, _ya, _za, _Aa, _Ba, _Ca, _Da, _Ea, _Fa, _Ga, _Ha, _Ia, _Ja, _Ka, _La, _Ma, _Na, _Oa, _Pa, _Qa, _Ra, _Sa, _Ta, _Ua, _Va, _Wa, _Xa, _Ya, _Za, __a, _$a, _ab, _bb, _cb, _db, _eb, _fb, _gb, _hb, _ib, _jb, _kb, _lb, _mb, _nb, _ob, _pb, _qb, _rb, _sb, _tb, _ub, _vb, _wb, _xb, _yb, _zb, _Ab, _Bb, _Cb, _Db, _Eb, _Fb, _Gb, _Hb, _Ib, _Jb, _Kb;
2931
+ var _a3, _b, _c, _d, _e, _f, _g2, _h, _i, _j, _k, _l, _m, _n, _o, _p, _q, _r, _s, _t, _u, _v, _w, _x, _y, _z, _A, _B, _C, _D, _E, _F, _G, _H, _I, _J, _K, _L, _M, _N, _O, _P, _Q, _R, _S, _T, _U, _V, _W, _X, _Y, _Z, __, _$, _aa, _ba, _ca, _da, _ea, _fa, _ga, _ha, _ia, _ja, _ka, _la, _ma, _na, _oa, _pa, _qa, _ra, _sa, _ta, _ua, _va, _wa, _xa, _ya, _za, _Aa, _Ba, _Ca, _Da, _Ea, _Fa, _Ga, _Ha, _Ia, _Ja, _Ka, _La, _Ma, _Na, _Oa, _Pa, _Qa, _Ra, _Sa, _Ta, _Ua, _Va, _Wa, _Xa, _Ya, _Za, __a, _$a, _ab, _bb, _cb, _db, _eb, _fb, _gb, _hb, _ib, _jb, _kb, _lb, _mb, _nb, _ob, _pb, _qb, _rb, _sb, _tb, _ub, _vb, _wb, _xb, _yb, _zb, _Ab, _Bb, _Cb, _Db, _Eb, _Fb, _Gb, _Hb;
2931
2932
  const licenseKey = (_d = (_c = (_b = (_a3 = readString(env, "RAG_LICENSE_KEY")) != null ? _a3 : readString(env, "RETRIVORA_LICENSE_KEY")) != null ? _b : readString(env, "NEXT_PUBLIC_RETRIVORA_LICENSE_KEY")) != null ? _c : readString(env, "LICENSE_KEY")) != null ? _d : base == null ? void 0 : base.licenseKey;
2932
2933
  const jwtProjectId = (() => {
2933
2934
  if (!licenseKey) return void 0;
@@ -3019,25 +3020,35 @@ function getEnvConfig(env = process.env, base) {
3019
3020
  return true;
3020
3021
  }
3021
3022
  })();
3022
- const defaultGatewayUrl = (_ra = (_qa = readString(env, "LITELLM_BASE_URL")) != null ? _qa : readString(env, "LLM_BASE_URL")) != null ? _ra : "https://www.retrivora.com/api/v1";
3023
- const defaultLlmProvider = isFreeTier2 ? "universal_rest" : "universal_rest";
3024
- const llmProvider = (_ua = (_ta = readEnum(env, "LLM_PROVIDER", defaultLlmProvider, LLM_PROVIDERS)) != null ? _ta : (_sa = base == null ? void 0 : base.llm) == null ? void 0 : _sa.provider) != null ? _ua : defaultLlmProvider;
3025
- const llmBaseUrl = (_ya = (_xa = (_va = readString(env, "LITELLM_BASE_URL")) != null ? _va : readString(env, "LLM_BASE_URL")) != null ? _xa : (_wa = base == null ? void 0 : base.llm) == null ? void 0 : _wa.baseUrl) != null ? _ya : defaultGatewayUrl;
3026
- const llmModel = (_Ca = (_Aa = readString(env, "LLM_MODEL")) != null ? _Aa : (_za = base == null ? void 0 : base.llm) == null ? void 0 : _za.model) != null ? _Ca : isFreeTier2 ? "llama-3.1-8b-instant" : (_Ba = DEFAULT_MODEL_BY_PROVIDER[llmProvider]) != null ? _Ba : "gpt-4o";
3027
- const llmApiKey = (_Ia = (_Ha = (_Fa = (_Ea = (_Da = llmApiKeyByProvider[llmProvider]) != null ? _Da : readString(env, "LLM_API_KEY")) != null ? _Ea : readString(env, "LITELLM_MASTER_KEY")) != null ? _Fa : readString(env, "LITELLM_API_KEY")) != null ? _Ha : (_Ga = base == null ? void 0 : base.llm) == null ? void 0 : _Ga.apiKey) != null ? _Ia : licenseKey;
3028
- const llmProfile = (_Ma = (_La = readString(env, "LLM_UNIVERSAL_PROFILE")) != null ? _La : (_Ka = (_Ja = base == null ? void 0 : base.llm) == null ? void 0 : _Ja.options) == null ? void 0 : _Ka.profile) != null ? _Ma : "litellm";
3029
- const defaultEmbeddingProvider = isFreeTier2 ? "universal_rest" : "universal_rest";
3030
- const embeddingProvider = (_Pa = (_Oa = readEnum(env, "EMBEDDING_PROVIDER", defaultEmbeddingProvider, EMBEDDING_PROVIDERS)) != null ? _Oa : (_Na = base == null ? void 0 : base.embedding) == null ? void 0 : _Na.provider) != null ? _Pa : defaultEmbeddingProvider;
3031
- const embeddingBaseUrl = (_Ua = (_Ta = (_Ra = (_Qa = readString(env, "LITELLM_BASE_URL")) != null ? _Qa : readString(env, "EMBEDDING_BASE_URL")) != null ? _Ra : readString(env, "LLM_BASE_URL")) != null ? _Ta : (_Sa = base == null ? void 0 : base.embedding) == null ? void 0 : _Sa.baseUrl) != null ? _Ua : defaultGatewayUrl;
3032
- const embeddingModel = (_Xa = (_Wa = readString(env, "EMBEDDING_MODEL")) != null ? _Wa : (_Va = base == null ? void 0 : base.embedding) == null ? void 0 : _Va.model) != null ? _Xa : "text-embedding-004";
3033
- const embeddingApiKey = (_cb = (_bb = (_$a = (__a = (_Za = (_Ya = embeddingApiKeyByProvider[embeddingProvider]) != null ? _Ya : readString(env, "EMBEDDING_API_KEY")) != null ? _Za : readString(env, "LLM_API_KEY")) != null ? __a : readString(env, "LITELLM_MASTER_KEY")) != null ? _$a : readString(env, "LITELLM_API_KEY")) != null ? _bb : (_ab = base == null ? void 0 : base.embedding) == null ? void 0 : _ab.apiKey) != null ? _cb : licenseKey;
3034
- const embeddingProfile = (_gb = (_fb = readString(env, "EMBEDDING_UNIVERSAL_PROFILE")) != null ? _fb : (_eb = (_db = base == null ? void 0 : base.embedding) == null ? void 0 : _db.options) == null ? void 0 : _eb.profile) != null ? _gb : "litellm";
3023
+ const tierBasedLLMModel = isFreeTier2 ? "llama-3.1-8b-instant" : "llama-3.3-70b-versatile";
3024
+ const tierBasedEmbeddingModel = isFreeTier2 ? "text-embedding-004" : "text-embedding-004";
3025
+ const defaultGatewayUrl = "https://llm.retrivora.com/api/v1";
3026
+ const rt = _runtimeConfig;
3027
+ const defaultLlmProvider = "universal_rest";
3028
+ const llmProvider = (_sa = (_ra = readEnum(env, "LLM_PROVIDER", defaultLlmProvider, LLM_PROVIDERS)) != null ? _ra : (_qa = base == null ? void 0 : base.llm) == null ? void 0 : _qa.provider) != null ? _sa : defaultLlmProvider;
3029
+ const llmBaseUrl = (_xa = (_wa = (_va = (_ta = readString(env, "LITELLM_BASE_URL")) != null ? _ta : readString(env, "LLM_BASE_URL")) != null ? _va : (_ua = base == null ? void 0 : base.llm) == null ? void 0 : _ua.baseUrl) != null ? _wa : rt == null ? void 0 : rt.llmBaseUrl) != null ? _xa : defaultGatewayUrl;
3030
+ const llmModel = (_Ba = (_Aa = (_za = readString(env, "LLM_MODEL")) != null ? _za : (_ya = base == null ? void 0 : base.llm) == null ? void 0 : _ya.model) != null ? _Aa : rt == null ? void 0 : rt.llmModel) != null ? _Ba : tierBasedLLMModel;
3031
+ const llmApiKey = (_Fa = (_Ea = (_Ca = readString(env, "LITELLM_MASTER_KEY")) != null ? _Ca : rt == null ? void 0 : rt.litellmApiKey) != null ? _Ea : (_Da = base == null ? void 0 : base.llm) == null ? void 0 : _Da.apiKey) != null ? _Fa : licenseKey;
3032
+ const llmProfile = (_Ja = (_Ia = readString(env, "LLM_UNIVERSAL_PROFILE")) != null ? _Ia : (_Ha = (_Ga = base == null ? void 0 : base.llm) == null ? void 0 : _Ga.options) == null ? void 0 : _Ha.profile) != null ? _Ja : "litellm";
3033
+ console.log("[serverConfig] \u{1F527} LLM Configuration:", {
3034
+ baseUrl: llmBaseUrl,
3035
+ model: llmModel,
3036
+ provider: llmProvider,
3037
+ hasApiKey: !!llmApiKey,
3038
+ source: (rt == null ? void 0 : rt.llmModel) ? "runtime-injected" : "env/fallback"
3039
+ });
3040
+ const defaultEmbeddingProvider = "universal_rest";
3041
+ const embeddingProvider = (_Ma = (_La = readEnum(env, "EMBEDDING_PROVIDER", defaultEmbeddingProvider, EMBEDDING_PROVIDERS)) != null ? _La : (_Ka = base == null ? void 0 : base.embedding) == null ? void 0 : _Ka.provider) != null ? _Ma : defaultEmbeddingProvider;
3042
+ const embeddingBaseUrl = (_Ta = (_Sa = (_Ra = (_Qa = (_Oa = (_Na = readString(env, "LITELLM_BASE_URL")) != null ? _Na : readString(env, "EMBEDDING_BASE_URL")) != null ? _Oa : readString(env, "LLM_BASE_URL")) != null ? _Qa : (_Pa = base == null ? void 0 : base.embedding) == null ? void 0 : _Pa.baseUrl) != null ? _Ra : rt == null ? void 0 : rt.embeddingBaseUrl) != null ? _Sa : rt == null ? void 0 : rt.llmBaseUrl) != null ? _Ta : defaultGatewayUrl;
3043
+ const embeddingModel = (_Xa = (_Wa = (_Va = readString(env, "EMBEDDING_MODEL")) != null ? _Va : (_Ua = base == null ? void 0 : base.embedding) == null ? void 0 : _Ua.model) != null ? _Wa : rt == null ? void 0 : rt.embeddingModel) != null ? _Xa : tierBasedEmbeddingModel;
3044
+ const embeddingApiKey = (_$a = (__a = (_Ya = readString(env, "LITELLM_MASTER_KEY")) != null ? _Ya : rt == null ? void 0 : rt.litellmApiKey) != null ? __a : (_Za = base == null ? void 0 : base.embedding) == null ? void 0 : _Za.apiKey) != null ? _$a : licenseKey;
3045
+ const embeddingProfile = (_db = (_cb = readString(env, "EMBEDDING_UNIVERSAL_PROFILE")) != null ? _cb : (_bb = (_ab = base == null ? void 0 : base.embedding) == null ? void 0 : _ab.options) == null ? void 0 : _bb.profile) != null ? _db : "litellm";
3035
3046
  return __spreadProps(__spreadValues({
3036
3047
  projectId,
3037
3048
  licenseKey,
3038
3049
  vectorDb: {
3039
3050
  provider: vectorProvider,
3040
- indexName: (_ib = (_hb = readString(env, "VECTOR_DB_INDEX")) != null ? _hb : vectorDbOptions.indexName) != null ? _ib : "rag-index",
3051
+ indexName: (_fb = (_eb = readString(env, "VECTOR_DB_INDEX")) != null ? _eb : vectorDbOptions.indexName) != null ? _fb : "rag-index",
3041
3052
  options: vectorDbOptions
3042
3053
  },
3043
3054
  llm: {
@@ -3067,30 +3078,30 @@ function getEnvConfig(env = process.env, base) {
3067
3078
  }
3068
3079
  },
3069
3080
  ui: {
3070
- title: (_kb = (_jb = readString(env, "NEXT_PUBLIC_UI_TITLE")) != null ? _jb : readString(env, "UI_TITLE")) != null ? _kb : "AI Assistant",
3071
- subtitle: (_mb = (_lb = readString(env, "NEXT_PUBLIC_UI_SUBTITLE")) != null ? _lb : readString(env, "UI_SUBTITLE")) != null ? _mb : "Powered by RAG",
3072
- primaryColor: (_ob = (_nb = readString(env, "NEXT_PUBLIC_PRIMARY_COLOR")) != null ? _nb : readString(env, "UI_PRIMARY_COLOR")) != null ? _ob : "#10b981",
3073
- accentColor: (_qb = (_pb = readString(env, "NEXT_PUBLIC_ACCENT_COLOR")) != null ? _pb : readString(env, "UI_ACCENT_COLOR")) != null ? _qb : "#3b82f6",
3074
- logoUrl: (_rb = readString(env, "NEXT_PUBLIC_LOGO_URL")) != null ? _rb : readString(env, "UI_LOGO_URL"),
3075
- placeholder: (_tb = (_sb = readString(env, "NEXT_PUBLIC_PLACEHOLDER")) != null ? _sb : readString(env, "UI_PLACEHOLDER")) != null ? _tb : "Ask me anything\u2026",
3076
- showSources: ((_vb = (_ub = readString(env, "NEXT_PUBLIC_SHOW_SOURCES")) != null ? _ub : readString(env, "UI_SHOW_SOURCES")) != null ? _vb : "true") !== "false",
3077
- welcomeMessage: (_xb = (_wb = readString(env, "NEXT_PUBLIC_WELCOME_MESSAGE")) != null ? _wb : readString(env, "UI_WELCOME_MESSAGE")) != null ? _xb : "Hello! I'm your AI assistant. Ask me anything about your documents.",
3078
- visualStyle: (_zb = (_yb = readString(env, "NEXT_PUBLIC_UI_VISUAL_STYLE")) != null ? _yb : readString(env, "UI_VISUAL_STYLE")) != null ? _zb : "glass",
3079
- borderRadius: (_Bb = (_Ab = readString(env, "NEXT_PUBLIC_UI_BORDER_RADIUS")) != null ? _Ab : readString(env, "UI_BORDER_RADIUS")) != null ? _Bb : "xl",
3080
- allowUpload: ((_Db = (_Cb = readString(env, "NEXT_PUBLIC_ALLOW_UPLOAD")) != null ? _Cb : readString(env, "UI_ALLOW_UPLOAD")) != null ? _Db : "false") === "true"
3081
+ title: (_hb = (_gb = readString(env, "NEXT_PUBLIC_UI_TITLE")) != null ? _gb : readString(env, "UI_TITLE")) != null ? _hb : "AI Assistant",
3082
+ subtitle: (_jb = (_ib = readString(env, "NEXT_PUBLIC_UI_SUBTITLE")) != null ? _ib : readString(env, "UI_SUBTITLE")) != null ? _jb : "Powered by RAG",
3083
+ primaryColor: (_lb = (_kb = readString(env, "NEXT_PUBLIC_PRIMARY_COLOR")) != null ? _kb : readString(env, "UI_PRIMARY_COLOR")) != null ? _lb : "#10b981",
3084
+ accentColor: (_nb = (_mb = readString(env, "NEXT_PUBLIC_ACCENT_COLOR")) != null ? _mb : readString(env, "UI_ACCENT_COLOR")) != null ? _nb : "#3b82f6",
3085
+ logoUrl: (_ob = readString(env, "NEXT_PUBLIC_LOGO_URL")) != null ? _ob : readString(env, "UI_LOGO_URL"),
3086
+ placeholder: (_qb = (_pb = readString(env, "NEXT_PUBLIC_PLACEHOLDER")) != null ? _pb : readString(env, "UI_PLACEHOLDER")) != null ? _qb : "Ask me anything\u2026",
3087
+ showSources: ((_sb = (_rb = readString(env, "NEXT_PUBLIC_SHOW_SOURCES")) != null ? _rb : readString(env, "UI_SHOW_SOURCES")) != null ? _sb : "true") !== "false",
3088
+ welcomeMessage: (_ub = (_tb = readString(env, "NEXT_PUBLIC_WELCOME_MESSAGE")) != null ? _tb : readString(env, "UI_WELCOME_MESSAGE")) != null ? _ub : "Hello! I'm your AI assistant. Ask me anything about your documents.",
3089
+ visualStyle: (_wb = (_vb = readString(env, "NEXT_PUBLIC_UI_VISUAL_STYLE")) != null ? _vb : readString(env, "UI_VISUAL_STYLE")) != null ? _wb : "glass",
3090
+ borderRadius: (_yb = (_xb = readString(env, "NEXT_PUBLIC_UI_BORDER_RADIUS")) != null ? _xb : readString(env, "UI_BORDER_RADIUS")) != null ? _yb : "xl",
3091
+ allowUpload: ((_Ab = (_zb = readString(env, "NEXT_PUBLIC_ALLOW_UPLOAD")) != null ? _zb : readString(env, "UI_ALLOW_UPLOAD")) != null ? _Ab : "false") === "true"
3081
3092
  },
3082
3093
  rag: {
3083
3094
  topK: readNumber(env, "RAG_TOP_K", 5),
3084
3095
  scoreThreshold: readNumber(env, "RAG_SCORE_THRESHOLD", 0),
3085
3096
  chunkSize: readNumber(env, "RAG_CHUNK_SIZE", 1e3),
3086
3097
  chunkOverlap: readNumber(env, "RAG_CHUNK_OVERLAP", 200),
3087
- filterableFields: (_Eb = readString(env, "RAG_FILTERABLE_FIELDS")) == null ? void 0 : _Eb.split(",").map((f) => f.trim()),
3098
+ filterableFields: (_Bb = readString(env, "RAG_FILTERABLE_FIELDS")) == null ? void 0 : _Bb.split(",").map((f) => f.trim()),
3088
3099
  // Query pipeline toggles — read from .env.local
3089
3100
  useQueryTransformation: readString(env, "RAG_USE_QUERY_TRANSFORMATION") === "true",
3090
3101
  useReranking: readString(env, "RAG_USE_RERANKING") === "true",
3091
3102
  useGraphRetrieval: readString(env, "RAG_USE_GRAPH_RETRIEVAL") === "true",
3092
- architecture: (_Fb = readString(env, "RAG_ARCHITECTURE")) != null ? _Fb : "simple",
3093
- chunkingStrategy: (_Gb = readString(env, "RAG_CHUNKING_STRATEGY")) != null ? _Gb : "recursive",
3103
+ architecture: (_Cb = readString(env, "RAG_ARCHITECTURE")) != null ? _Cb : "simple",
3104
+ chunkingStrategy: (_Db = readString(env, "RAG_CHUNKING_STRATEGY")) != null ? _Db : "recursive",
3094
3105
  uiMapping: (() => {
3095
3106
  const raw = readString(env, "RAG_UI_MAPPING");
3096
3107
  if (!raw) return void 0;
@@ -3103,7 +3114,7 @@ function getEnvConfig(env = process.env, base) {
3103
3114
  },
3104
3115
  telemetry: {
3105
3116
  enabled: telemetryEnabled,
3106
- url: (_Kb = (_Jb = (_Hb = readString(env, "TELEMETRY_URL")) != null ? _Hb : readString(env, "NEXT_PUBLIC_TELEMETRY_URL")) != null ? _Jb : (_Ib = base == null ? void 0 : base.telemetry) == null ? void 0 : _Ib.url) != null ? _Kb : "https://www.retrivora.com/api/telemetry"
3117
+ url: (_Hb = (_Gb = (_Eb = readString(env, "TELEMETRY_URL")) != null ? _Eb : readString(env, "NEXT_PUBLIC_TELEMETRY_URL")) != null ? _Gb : (_Fb = base == null ? void 0 : base.telemetry) == null ? void 0 : _Fb.url) != null ? _Hb : "https://www.retrivora.com/api/telemetry"
3107
3118
  }
3108
3119
  }, readString(env, "GRAPH_DB_PROVIDER") ? {
3109
3120
  graphDb: {
@@ -4421,7 +4432,7 @@ var UniversalLLMAdapter = class {
4421
4432
  });
4422
4433
  }
4423
4434
  async chat(messages, context) {
4424
- var _a3, _b, _c, _d, _e;
4435
+ var _a3, _b, _c, _d, _e, _f, _g2, _h, _i, _j;
4425
4436
  const path2 = (_a3 = this.opts.chatPath) != null ? _a3 : "/chat/completions";
4426
4437
  const safeMessages = (Array.isArray(messages) ? messages : []).filter((m) => Boolean(m && typeof m === "object")).map((m) => ({
4427
4438
  role: m.role || "user",
@@ -4451,16 +4462,27 @@ ${context != null ? context : "None"}` },
4451
4462
  };
4452
4463
  }
4453
4464
  try {
4465
+ console.debug(`[UniversalLLMAdapter] POST ${this.baseUrl}${path2}`);
4466
+ console.debug(`[UniversalLLMAdapter] Model: ${this.model}`);
4467
+ console.debug(`[UniversalLLMAdapter] Auth: Bearer ${((_b = this.apiKey) == null ? void 0 : _b.substring(0, 20)) || "none"}...`);
4454
4468
  const { data: data2 } = await this.http.post(path2, payload);
4455
- const extractPath2 = (_b = this.opts.responseExtractPath) != null ? _b : "choices[0].message.content";
4456
- const result2 = (_c = resolvePath(data2, extractPath2)) != null ? _c : extractContent(data2);
4469
+ const extractPath2 = (_c = this.opts.responseExtractPath) != null ? _c : "choices[0].message.content";
4470
+ const result2 = (_d = resolvePath(data2, extractPath2)) != null ? _d : extractContent(data2);
4457
4471
  if (result2 !== void 0) {
4458
4472
  return String(result2);
4459
4473
  }
4460
4474
  } catch (httpErr) {
4461
- const _g2 = globalThis;
4462
- const dispatch = _g2.__retrivoraDispatchChat;
4475
+ console.error("[UniversalLLMAdapter] \u274C HTTP chat request failed:", {
4476
+ status: (_e = httpErr == null ? void 0 : httpErr.response) == null ? void 0 : _e.status,
4477
+ statusText: (_f = httpErr == null ? void 0 : httpErr.response) == null ? void 0 : _f.statusText,
4478
+ url: (_g2 = httpErr == null ? void 0 : httpErr.config) == null ? void 0 : _g2.url,
4479
+ errorData: (_h = httpErr == null ? void 0 : httpErr.response) == null ? void 0 : _h.data,
4480
+ message: httpErr == null ? void 0 : httpErr.message
4481
+ });
4482
+ const _g3 = globalThis;
4483
+ const dispatch = _g3.__retrivoraDispatchChat;
4463
4484
  if (typeof dispatch === "function") {
4485
+ console.debug("[UniversalLLMAdapter] \u{1F504} Attempting in-process fallback dispatch");
4464
4486
  const res = await dispatch({
4465
4487
  model: this.model,
4466
4488
  messages: formattedMessages,
@@ -4475,8 +4497,8 @@ ${context != null ? context : "None"}` },
4475
4497
  throw httpErr;
4476
4498
  }
4477
4499
  const { data } = await this.http.post(path2, payload);
4478
- const extractPath = (_d = this.opts.responseExtractPath) != null ? _d : "choices[0].message.content";
4479
- const result = (_e = resolvePath(data, extractPath)) != null ? _e : extractContent(data);
4500
+ const extractPath = (_i = this.opts.responseExtractPath) != null ? _i : "choices[0].message.content";
4501
+ const result = (_j = resolvePath(data, extractPath)) != null ? _j : extractContent(data);
4480
4502
  if (result === void 0) {
4481
4503
  throw new Error(`[UniversalLLMAdapter] Could not extract text from path '${extractPath}' in response.`);
4482
4504
  }
@@ -4615,36 +4637,70 @@ ${context != null ? context : "None"}` },
4615
4637
  });
4616
4638
  }
4617
4639
  async embed(text) {
4618
- var _a3, _b, _c, _d, _e;
4640
+ var _a3, _b, _c, _d, _e, _f, _g2;
4619
4641
  const path2 = (_a3 = this.opts.embedPath) != null ? _a3 : "/embeddings";
4620
4642
  const payload = this.opts.embedPayloadTemplate ? buildPayload(this.opts.embedPayloadTemplate, { input: text, model: this.model }) : { input: text, model: this.model };
4621
4643
  try {
4622
- const { data: data2 } = await this.http.post(path2, payload);
4623
- const extractPath2 = (_b = this.opts.embedExtractPath) != null ? _b : "data[0].embedding";
4624
- const vector2 = resolvePath(data2, extractPath2);
4625
- if (Array.isArray(vector2)) return vector2;
4626
- } catch (e) {
4627
- const _g2 = globalThis;
4628
- const dispatchEmbed = _g2.__retrivoraDispatchEmbedding;
4644
+ const { data } = await this.http.post(path2, payload);
4645
+ const extractPath = (_b = this.opts.embedExtractPath) != null ? _b : "data[0].embedding";
4646
+ const vector = resolvePath(data, extractPath);
4647
+ if (Array.isArray(vector)) return vector;
4648
+ console.warn(
4649
+ `[UniversalLLMAdapter] Embedding API returned unexpected format for path '${extractPath}'. Response:`,
4650
+ data
4651
+ );
4652
+ } catch (err) {
4653
+ console.warn(
4654
+ `[UniversalLLMAdapter] HTTP embedding request failed: ${(err == null ? void 0 : err.message) || err}`,
4655
+ { path: path2, model: this.model }
4656
+ );
4657
+ const _g3 = globalThis;
4658
+ const dispatchEmbed = _g3.__retrivoraDispatchEmbedding;
4629
4659
  if (typeof dispatchEmbed === "function") {
4630
- const res = await dispatchEmbed({ input: text, model: this.model }, this.apiKey);
4631
- if ((_d = (_c = res == null ? void 0 : res.data) == null ? void 0 : _c[0]) == null ? void 0 : _d.embedding) {
4632
- return res.data[0].embedding;
4660
+ try {
4661
+ const res = await dispatchEmbed({ input: text, model: this.model }, this.apiKey);
4662
+ if ((_d = (_c = res == null ? void 0 : res.data) == null ? void 0 : _c[0]) == null ? void 0 : _d.embedding) {
4663
+ return res.data[0].embedding;
4664
+ }
4665
+ } catch (dispatchErr) {
4666
+ console.warn("[UniversalLLMAdapter] In-process dispatch embedding failed:", dispatchErr);
4633
4667
  }
4634
4668
  }
4635
4669
  }
4636
- const { data } = await this.http.post(path2, payload);
4637
- const extractPath = (_e = this.opts.embedExtractPath) != null ? _e : "data[0].embedding";
4638
- const vector = resolvePath(data, extractPath);
4639
- if (!Array.isArray(vector)) {
4640
- throw new Error(`[UniversalLLMAdapter] Could not extract embedding vector from path '${extractPath}' in response.`);
4670
+ try {
4671
+ const { data } = await this.http.post(path2, payload);
4672
+ const extractPath = (_e = this.opts.embedExtractPath) != null ? _e : "data[0].embedding";
4673
+ const vector = resolvePath(data, extractPath);
4674
+ if (Array.isArray(vector)) {
4675
+ return vector;
4676
+ }
4677
+ console.error(
4678
+ `[UniversalLLMAdapter] Retry embedding failed: unexpected response format for path '${extractPath}'. Response:`,
4679
+ data
4680
+ );
4681
+ throw new Error(
4682
+ `[UniversalLLMAdapter] Could not extract embedding vector. Expected array at '${extractPath}', got: ${typeof vector}. Response keys: ${Object.keys(data || {})}.`
4683
+ );
4684
+ } catch (err) {
4685
+ console.error(
4686
+ `[UniversalLLMAdapter] Both embedding attempts failed. Path: ${path2}, Model: ${this.model}`,
4687
+ { error: err == null ? void 0 : err.message, statusCode: (_f = err == null ? void 0 : err.response) == null ? void 0 : _f.status, response: (_g2 = err == null ? void 0 : err.response) == null ? void 0 : _g2.data }
4688
+ );
4689
+ throw new Error(
4690
+ `[UniversalLLMAdapter] Failed to generate embedding after 2 attempts: ${(err == null ? void 0 : err.message) || err}`
4691
+ );
4641
4692
  }
4642
- return vector;
4643
4693
  }
4644
4694
  async batchEmbed(texts) {
4645
4695
  const results = [];
4646
4696
  for (const text of texts) {
4647
- results.push(await this.embed(text));
4697
+ try {
4698
+ const vector = await this.embed(text);
4699
+ results.push(vector);
4700
+ } catch (err) {
4701
+ console.warn(`[UniversalLLMAdapter] batchEmbed failed for text (first 50 chars): ${text.slice(0, 50)}...`, err);
4702
+ throw err;
4703
+ }
4648
4704
  }
4649
4705
  return results;
4650
4706
  }
@@ -11377,6 +11433,7 @@ function createUploadHandler(configOrPlugin, options) {
11377
11433
  const plugin = getOrCreatePlugin(configOrPlugin);
11378
11434
  const onAuthorize = options == null ? void 0 : options.onAuthorize;
11379
11435
  return async function POST(req, context) {
11436
+ console.log("Upload handler invoked");
11380
11437
  const authResult = await checkAuth(req, onAuthorize);
11381
11438
  if (authResult) return authResult;
11382
11439
  const config = plugin.getConfig();
@@ -11385,6 +11442,7 @@ function createUploadHandler(configOrPlugin, options) {
11385
11442
  try {
11386
11443
  const formData = await req.formData();
11387
11444
  const files = formData.getAll("files");
11445
+ console.log("Files received for upload:", files.map((f) => f.name));
11388
11446
  const namespace = formData.get("namespace") || void 0;
11389
11447
  const dimensionRaw = formData.get("dimension");
11390
11448
  const dimension = dimensionRaw ? parseInt(dimensionRaw, 10) : void 0;