@retrivora-ai/rag-engine 0.1.0

package/dist/server.d.mts

@@ -0,0 +1,211 @@
+import { f as RagConfig, h as VectorDBConfig, d as IVectorDB, g as UpsertDocument, V as VectorMatch, L as LLMConfig, c as EmbeddingConfig, I as ILLMProvider, C as ChatMessage, a as ChatOptions, E as EmbedOptions } from './RAGPipeline-BmkIv1HD.mjs';
+export { b as ChatResponse, e as IngestDocument, R as RAGConfig, i as RAGPipeline, U as UIConfig } from './RAGPipeline-BmkIv1HD.mjs';
+export { C as Chunk, a as ChunkOptions, D as DocumentChunker } from './DocumentChunker-BUrIrcPk.mjs';
+export { createChatHandler, createHealthHandler, createIngestHandler, createUploadHandler } from './handlers/index.mjs';
+import 'next/server';
+
+/**
+ * serverConfig.ts — reads RagConfig from environment variables at runtime.
+ *
+ * This keeps secrets server-side and never exposes them to the browser.
+ * Consumer projects set these env vars in their .env.local file.
+ */
+
+declare function getRagConfig(env?: Record<string, string | undefined>): RagConfig;
+
+/**
+ * VectorDBFactory — instantiates the correct IVectorDB adapter
+ * based on the VectorDBConfig.provider value.
+ *
+ * Usage:
+ *   const db = VectorDBFactory.create(config.vectorDb);
+ *   await db.initialize();
+ */
+
+declare class VectorDBFactory {
+    static create(config: VectorDBConfig): IVectorDB;
+}
+
+/**
+ * Pinecone Vector DB Adapter
+ *
+ * Required options in VectorDBConfig.options:
+ *   - apiKey:       string  – Pinecone API key
+ *   - environment?: string  – Pinecone environment (legacy, optional for new projects)
+ */
+
+declare class PineconeAdapter implements IVectorDB {
+    private client;
+    private readonly indexName;
+    private readonly apiKey;
+    constructor(config: VectorDBConfig);
+    initialize(): Promise<void>;
+    private index;
+    upsert(doc: UpsertDocument, namespace?: string): Promise<void>;
+    batchUpsert(docs: UpsertDocument[], namespace?: string): Promise<void>;
+    query(vector: number[], topK: number, namespace?: string, filter?: Record<string, unknown>): Promise<VectorMatch[]>;
+    delete(id: string, namespace?: string): Promise<void>;
+    deleteNamespace(namespace: string): Promise<void>;
+    ping(): Promise<boolean>;
+    disconnect(): Promise<void>;
+}
+
+/**
+ * pgVector Adapter
+ *
+ * Stores embeddings in a PostgreSQL table using the pgvector extension.
+ * Required options in VectorDBConfig.options:
+ *   - connectionString: string  – PostgreSQL connection string
+ *   - dimensions?:      number  – vector dimensions (default 1536 for OpenAI ada-002)
+ *
+ * The adapter auto-creates the table and index on initialize().
+ */
+
+declare class PgVectorAdapter implements IVectorDB {
+    private pool;
+    private readonly tableName;
+    private readonly dimensions;
+    private readonly connectionString;
+    constructor(config: VectorDBConfig);
+    initialize(): Promise<void>;
+    upsert(doc: UpsertDocument, namespace?: string): Promise<void>;
+    batchUpsert(docs: UpsertDocument[], namespace?: string): Promise<void>;
+    query(vector: number[], topK: number, namespace?: string, filter?: Record<string, unknown>): Promise<VectorMatch[]>;
+    delete(id: string, namespace?: string): Promise<void>;
+    deleteNamespace(namespace: string): Promise<void>;
+    ping(): Promise<boolean>;
+    disconnect(): Promise<void>;
+}
+
+/**
+ * MongoDB Adapter for Atlas Vector Search.
+ *
+ * Requirements:
+ * 1. A MongoDB Atlas cluster (v6.0.11+, v7.0.2+).
+ * 2. A Vector Search Index created on the collection.
+ *
+ * Config Options:
+ * - uri: string (Required)
+ * - database: string (Required)
+ * - collection: string (Required)
+ * - indexName: string (Default: 'vector_index')
+ * - embeddingKey: string (Default: 'embedding')
+ * - contentKey: string (Default: 'content')
+ * - metadataKey: string (Default: 'metadata')
+ */
+declare class MongoDbAdapter implements IVectorDB {
+    private client;
+    private db?;
+    private collection?;
+    private uri;
+    private dbName;
+    private collectionName;
+    private indexName;
+    private embeddingKey;
+    private contentKey;
+    private metadataKey;
+    constructor(options: Record<string, unknown>);
+    initialize(): Promise<void>;
+    upsert(doc: UpsertDocument, namespace?: string): Promise<void>;
+    batchUpsert(docs: UpsertDocument[], namespace?: string): Promise<void>;
+    query(vector: number[], topK: number, namespace?: string, filter?: Record<string, unknown>): Promise<VectorMatch[]>;
+    delete(id: string, namespace?: string): Promise<void>;
+    deleteNamespace(namespace: string): Promise<void>;
+    ping(): Promise<boolean>;
+    disconnect(): Promise<void>;
+}
+
+/**
+ * LLMFactory — instantiates the correct ILLMProvider based on llmConfig.provider.
+ *
+ * Also accepts an optional EmbeddingConfig so providers that support embedding
+ * can be initialised with separate embedding credentials.
+ */
+
+declare class LLMFactory {
+    static create(llmConfig: LLMConfig, embeddingConfig?: EmbeddingConfig): ILLMProvider;
+    /**
+     * Creates a dedicated embedding-only provider.
+     * Useful when the LLM provider (e.g. Anthropic) doesn't support embeddings.
+     */
+    static createEmbeddingProvider(embeddingConfig: EmbeddingConfig): ILLMProvider;
+}
+
+/**
+ * OpenAI LLM Provider
+ *
+ * Handles both chat completion (GPT-4o, GPT-4-turbo, etc.) and
+ * embedding (text-embedding-ada-002, text-embedding-3-small/large).
+ *
+ * Required LLMConfig fields:
+ *   - apiKey: string  – OpenAI API key
+ *   - model:  string  – e.g. "gpt-4o"
+ *
+ * Required EmbeddingConfig fields (when used for embedding):
+ *   - apiKey: string  – same or separate key
+ *   - model:  string  – e.g. "text-embedding-3-small"
+ */
+
+declare class OpenAIProvider implements ILLMProvider {
+    private readonly client;
+    private readonly llmConfig;
+    private readonly embeddingConfig?;
+    constructor(llmConfig: LLMConfig, embeddingConfig?: EmbeddingConfig);
+    chat(messages: ChatMessage[], context: string, options?: ChatOptions): Promise<string>;
+    embed(text: string, options?: EmbedOptions): Promise<number[]>;
+    batchEmbed(texts: string[], options?: EmbedOptions): Promise<number[][]>;
+    ping(): Promise<boolean>;
+}
+
+/**
+ * Anthropic (Claude) LLM Provider
+ *
+ * Handles chat completion via Anthropic's Messages API.
+ * Note: Anthropic does NOT provide embedding models — use OpenAIProvider
+ * or OllamaProvider for embedding when pairing with this provider.
+ *
+ * Required LLMConfig fields:
+ *   - apiKey: string  – Anthropic API key
+ *   - model:  string  – e.g. "claude-3-5-sonnet-20241022"
+ */
+
+declare class AnthropicProvider implements ILLMProvider {
+    private readonly client;
+    private readonly llmConfig;
+    private readonly embeddingConfig?;
+    constructor(llmConfig: LLMConfig, embeddingConfig?: EmbeddingConfig);
+    chat(messages: ChatMessage[], context: string, options?: ChatOptions): Promise<string>;
+    /**
+     * Anthropic does not offer an embedding API.
+     * This method throws with a clear error so developers know to configure
+     * a separate embedding provider (OpenAI or Ollama).
+     */
+    embed(text: string, options?: EmbedOptions): Promise<number[]>;
+    batchEmbed(texts: string[], options?: EmbedOptions): Promise<number[][]>;
+    ping(): Promise<boolean>;
+}
+
+/**
+ * Ollama LLM Provider (local / self-hosted)
+ *
+ * Communicates with an Ollama server via its REST API.
+ * Supports both chat completion and embedding (requires a model with embed support,
+ * e.g. "nomic-embed-text", "mxbai-embed-large").
+ *
+ * Required LLMConfig fields:
+ *   - model:    string  – e.g. "llama3", "mistral", "gemma2"
+ *   - baseUrl?: string  – Ollama server URL (default "http://localhost:11434")
+ */
+
+declare class OllamaProvider implements ILLMProvider {
+    private readonly http;
+    private readonly llmConfig;
+    private readonly embeddingConfig?;
+    constructor(llmConfig: LLMConfig, embeddingConfig?: EmbeddingConfig);
+    chat(messages: ChatMessage[], context: string, options?: ChatOptions): Promise<string>;
+    embed(text: string, options?: EmbedOptions): Promise<number[]>;
+    batchEmbed(texts: string[], options?: EmbedOptions): Promise<number[][]>;
+    ping(): Promise<boolean>;
+}
+
+export { AnthropicProvider, ChatMessage, ChatOptions, EmbedOptions, EmbeddingConfig, ILLMProvider, IVectorDB, LLMConfig, LLMFactory, MongoDbAdapter, OllamaProvider, OpenAIProvider, PgVectorAdapter, PineconeAdapter, RagConfig, UpsertDocument, VectorDBConfig, VectorDBFactory, VectorMatch, getRagConfig };

package/dist/server.d.ts

@@ -0,0 +1,211 @@
+import { f as RagConfig, h as VectorDBConfig, d as IVectorDB, g as UpsertDocument, V as VectorMatch, L as LLMConfig, c as EmbeddingConfig, I as ILLMProvider, C as ChatMessage, a as ChatOptions, E as EmbedOptions } from './RAGPipeline-BmkIv1HD.js';
+export { b as ChatResponse, e as IngestDocument, R as RAGConfig, i as RAGPipeline, U as UIConfig } from './RAGPipeline-BmkIv1HD.js';
+export { C as Chunk, a as ChunkOptions, D as DocumentChunker } from './DocumentChunker-BUrIrcPk.js';
+export { createChatHandler, createHealthHandler, createIngestHandler, createUploadHandler } from './handlers/index.js';
+import 'next/server';
+
+/**
+ * serverConfig.ts — reads RagConfig from environment variables at runtime.
+ *
+ * This keeps secrets server-side and never exposes them to the browser.
+ * Consumer projects set these env vars in their .env.local file.
+ */
+
+declare function getRagConfig(env?: Record<string, string | undefined>): RagConfig;
+
+/**
+ * VectorDBFactory — instantiates the correct IVectorDB adapter
+ * based on the VectorDBConfig.provider value.
+ *
+ * Usage:
+ *   const db = VectorDBFactory.create(config.vectorDb);
+ *   await db.initialize();
+ */
+
+declare class VectorDBFactory {
+    static create(config: VectorDBConfig): IVectorDB;
+}
+
+/**
+ * Pinecone Vector DB Adapter
+ *
+ * Required options in VectorDBConfig.options:
+ *   - apiKey:       string  – Pinecone API key
+ *   - environment?: string  – Pinecone environment (legacy, optional for new projects)
+ */
+
+declare class PineconeAdapter implements IVectorDB {
+    private client;
+    private readonly indexName;
+    private readonly apiKey;
+    constructor(config: VectorDBConfig);
+    initialize(): Promise<void>;
+    private index;
+    upsert(doc: UpsertDocument, namespace?: string): Promise<void>;
+    batchUpsert(docs: UpsertDocument[], namespace?: string): Promise<void>;
+    query(vector: number[], topK: number, namespace?: string, filter?: Record<string, unknown>): Promise<VectorMatch[]>;
+    delete(id: string, namespace?: string): Promise<void>;
+    deleteNamespace(namespace: string): Promise<void>;
+    ping(): Promise<boolean>;
+    disconnect(): Promise<void>;
+}
+
+/**
+ * pgVector Adapter
+ *
+ * Stores embeddings in a PostgreSQL table using the pgvector extension.
+ * Required options in VectorDBConfig.options:
+ *   - connectionString: string  – PostgreSQL connection string
+ *   - dimensions?:      number  – vector dimensions (default 1536 for OpenAI ada-002)
+ *
+ * The adapter auto-creates the table and index on initialize().
+ */
+
+declare class PgVectorAdapter implements IVectorDB {
+    private pool;
+    private readonly tableName;
+    private readonly dimensions;
+    private readonly connectionString;
+    constructor(config: VectorDBConfig);
+    initialize(): Promise<void>;
+    upsert(doc: UpsertDocument, namespace?: string): Promise<void>;
+    batchUpsert(docs: UpsertDocument[], namespace?: string): Promise<void>;
+    query(vector: number[], topK: number, namespace?: string, filter?: Record<string, unknown>): Promise<VectorMatch[]>;
+    delete(id: string, namespace?: string): Promise<void>;
+    deleteNamespace(namespace: string): Promise<void>;
+    ping(): Promise<boolean>;
+    disconnect(): Promise<void>;
+}
+
+/**
+ * MongoDB Adapter for Atlas Vector Search.
+ *
+ * Requirements:
+ * 1. A MongoDB Atlas cluster (v6.0.11+, v7.0.2+).
+ * 2. A Vector Search Index created on the collection.
+ *
+ * Config Options:
+ * - uri: string (Required)
+ * - database: string (Required)
+ * - collection: string (Required)
+ * - indexName: string (Default: 'vector_index')
+ * - embeddingKey: string (Default: 'embedding')
+ * - contentKey: string (Default: 'content')
+ * - metadataKey: string (Default: 'metadata')
+ */
+declare class MongoDbAdapter implements IVectorDB {
+    private client;
+    private db?;
+    private collection?;
+    private uri;
+    private dbName;
+    private collectionName;
+    private indexName;
+    private embeddingKey;
+    private contentKey;
+    private metadataKey;
+    constructor(options: Record<string, unknown>);
+    initialize(): Promise<void>;
+    upsert(doc: UpsertDocument, namespace?: string): Promise<void>;
+    batchUpsert(docs: UpsertDocument[], namespace?: string): Promise<void>;
+    query(vector: number[], topK: number, namespace?: string, filter?: Record<string, unknown>): Promise<VectorMatch[]>;
+    delete(id: string, namespace?: string): Promise<void>;
+    deleteNamespace(namespace: string): Promise<void>;
+    ping(): Promise<boolean>;
+    disconnect(): Promise<void>;
+}
+
+/**
+ * LLMFactory — instantiates the correct ILLMProvider based on llmConfig.provider.
+ *
+ * Also accepts an optional EmbeddingConfig so providers that support embedding
+ * can be initialised with separate embedding credentials.
+ */
+
+declare class LLMFactory {
+    static create(llmConfig: LLMConfig, embeddingConfig?: EmbeddingConfig): ILLMProvider;
+    /**
+     * Creates a dedicated embedding-only provider.
+     * Useful when the LLM provider (e.g. Anthropic) doesn't support embeddings.
+     */
+    static createEmbeddingProvider(embeddingConfig: EmbeddingConfig): ILLMProvider;
+}
+
+/**
+ * OpenAI LLM Provider
+ *
+ * Handles both chat completion (GPT-4o, GPT-4-turbo, etc.) and
+ * embedding (text-embedding-ada-002, text-embedding-3-small/large).
+ *
+ * Required LLMConfig fields:
+ *   - apiKey: string  – OpenAI API key
+ *   - model:  string  – e.g. "gpt-4o"
+ *
+ * Required EmbeddingConfig fields (when used for embedding):
+ *   - apiKey: string  – same or separate key
+ *   - model:  string  – e.g. "text-embedding-3-small"
+ */
+
+declare class OpenAIProvider implements ILLMProvider {
+    private readonly client;
+    private readonly llmConfig;
+    private readonly embeddingConfig?;
+    constructor(llmConfig: LLMConfig, embeddingConfig?: EmbeddingConfig);
+    chat(messages: ChatMessage[], context: string, options?: ChatOptions): Promise<string>;
+    embed(text: string, options?: EmbedOptions): Promise<number[]>;
+    batchEmbed(texts: string[], options?: EmbedOptions): Promise<number[][]>;
+    ping(): Promise<boolean>;
+}
+
+/**
+ * Anthropic (Claude) LLM Provider
+ *
+ * Handles chat completion via Anthropic's Messages API.
+ * Note: Anthropic does NOT provide embedding models — use OpenAIProvider
+ * or OllamaProvider for embedding when pairing with this provider.
+ *
+ * Required LLMConfig fields:
+ *   - apiKey: string  – Anthropic API key
+ *   - model:  string  – e.g. "claude-3-5-sonnet-20241022"
+ */
+
+declare class AnthropicProvider implements ILLMProvider {
+    private readonly client;
+    private readonly llmConfig;
+    private readonly embeddingConfig?;
+    constructor(llmConfig: LLMConfig, embeddingConfig?: EmbeddingConfig);
+    chat(messages: ChatMessage[], context: string, options?: ChatOptions): Promise<string>;
+    /**
+     * Anthropic does not offer an embedding API.
+     * This method throws with a clear error so developers know to configure
+     * a separate embedding provider (OpenAI or Ollama).
+     */
+    embed(text: string, options?: EmbedOptions): Promise<number[]>;
+    batchEmbed(texts: string[], options?: EmbedOptions): Promise<number[][]>;
+    ping(): Promise<boolean>;
+}
+
+/**
+ * Ollama LLM Provider (local / self-hosted)
+ *
+ * Communicates with an Ollama server via its REST API.
+ * Supports both chat completion and embedding (requires a model with embed support,
+ * e.g. "nomic-embed-text", "mxbai-embed-large").
+ *
+ * Required LLMConfig fields:
+ *   - model:    string  – e.g. "llama3", "mistral", "gemma2"
+ *   - baseUrl?: string  – Ollama server URL (default "http://localhost:11434")
+ */
+
+declare class OllamaProvider implements ILLMProvider {
+    private readonly http;
+    private readonly llmConfig;
+    private readonly embeddingConfig?;
+    constructor(llmConfig: LLMConfig, embeddingConfig?: EmbeddingConfig);
+    chat(messages: ChatMessage[], context: string, options?: ChatOptions): Promise<string>;
+    embed(text: string, options?: EmbedOptions): Promise<number[]>;
+    batchEmbed(texts: string[], options?: EmbedOptions): Promise<number[][]>;
+    ping(): Promise<boolean>;
+}
+
+export { AnthropicProvider, ChatMessage, ChatOptions, EmbedOptions, EmbeddingConfig, ILLMProvider, IVectorDB, LLMConfig, LLMFactory, MongoDbAdapter, OllamaProvider, OpenAIProvider, PgVectorAdapter, PineconeAdapter, RagConfig, UpsertDocument, VectorDBConfig, VectorDBFactory, VectorMatch, getRagConfig };