@retrivora-ai/rag-engine 0.1.7 → 0.1.9

package/dist/server.mjs

@@ -1,70 +1,481 @@
-import {
-  QdrantProvider
-} from "./chunk-CWQQHAF6.mjs";
 import {
   ChromaDBProvider
-} from "./chunk-HUGLYKD6.mjs";
+} from "./chunk-VKE5ZW7Y.mjs";
 import {
   RedisProvider
-} from "./chunk-VPNRDXIA.mjs";
+} from "./chunk-HOMXEE3M.mjs";
 import {
   WeaviateProvider
-} from "./chunk-V75V7BT2.mjs";
+} from "./chunk-RUKZC3ON.mjs";
+import {
+  UniversalVectorProvider
+} from "./chunk-VV2ML6TM.mjs";
 import {
+  AnthropicProvider,
+  BatchProcessor,
   ConfigResolver,
+  ConfigValidator,
   DocumentChunker,
+  DocumentParser,
+  EmbeddingStrategy,
+  EmbeddingStrategyResolver,
+  LLMFactory,
+  LLM_PROFILES,
+  OllamaProvider,
+  OpenAIProvider,
   Pipeline,
+  ProviderHealthCheck,
   ProviderRegistry,
+  UniversalLLMAdapter,
+  VECTOR_PROFILES,
   VectorPlugin,
   createChatHandler,
   createHealthHandler,
   createIngestHandler,
   createUploadHandler,
   getRagConfig
-} from "./chunk-7YQWGERZ.mjs";
-import {
-  AnthropicProvider,
-  LLMFactory,
-  OllamaProvider,
-  OpenAIProvider
-} from "./chunk-JI6VD5TJ.mjs";
-import "./chunk-UKDXCXW7.mjs";
+} from "./chunk-2VR5ZMXV.mjs";
+import "./chunk-EDLTMSNY.mjs";
 import {
   PineconeProvider
-} from "./chunk-ZM6TYIDH.mjs";
+} from "./chunk-BMHJTWSU.mjs";
 import {
   PostgreSQLProvider
 } from "./chunk-IUTAZ7QR.mjs";
 import {
   MongoDBProvider
-} from "./chunk-QEYVWVT5.mjs";
+} from "./chunk-5HXNKSCR.mjs";
 import {
   MilvusProvider
-} from "./chunk-7NXI6ZWX.mjs";
+} from "./chunk-VEJNRS4B.mjs";
+import {
+  QdrantProvider
+} from "./chunk-W2PQR3UK.mjs";
 import {
   BaseVectorProvider
 } from "./chunk-VOIWNO5O.mjs";
-import "./chunk-I4E63NIC.mjs";
+import {
+  __objRest,
+  __spreadProps,
+  __spreadValues
+} from "./chunk-FWCSY2DS.mjs";
+
+// src/config/ConfigBuilder.ts
+var ConfigBuilder = class {
+  constructor() {
+    this.config = {
+      projectId: "default-project",
+      rag: {
+        chunkSize: 1e3,
+        chunkOverlap: 200,
+        topK: 5
+      }
+    };
+  }
+  /**
+   * Set the project/application ID for namespacing
+   */
+  projectId(id) {
+    this.config.projectId = id;
+    return this;
+  }
+  /**
+   * Configure the vector database provider
+   */
+  vectorDb(provider, options) {
+    var _a;
+    if (provider === "auto") {
+      this.config.vectorDb = this.autoDetectVectorDb();
+    } else {
+      this.config.vectorDb = {
+        provider,
+        indexName: (_a = options == null ? void 0 : options.indexName) != null ? _a : "default",
+        options: __spreadValues({}, options)
+      };
+    }
+    return this;
+  }
+  /**
+   * Configure the LLM provider for chat
+   */
+  llm(provider, model, apiKey, options) {
+    var _a, _b;
+    if (provider === "auto") {
+      this.config.llm = this.autoDetectLLM();
+    } else {
+      this.config.llm = {
+        provider,
+        model: model != null ? model : "default-model",
+        apiKey,
+        systemPrompt: options == null ? void 0 : options.systemPrompt,
+        maxTokens: (_a = options == null ? void 0 : options.maxTokens) != null ? _a : 1024,
+        temperature: (_b = options == null ? void 0 : options.temperature) != null ? _b : 0.7,
+        baseUrl: options == null ? void 0 : options.baseUrl,
+        options
+      };
+    }
+    return this;
+  }
+  /**
+   * Configure the embedding provider
+   */
+  embedding(provider, model, apiKey, options) {
+    if (provider === "auto") {
+      this.config.embedding = this.autoDetectEmbedding();
+    } else {
+      this.config.embedding = {
+        provider,
+        model: model != null ? model : "default-embedding",
+        apiKey,
+        baseUrl: options == null ? void 0 : options.baseUrl,
+        options
+      };
+    }
+    return this;
+  }
+  /**
+   * Set RAG-specific parameters
+   */
+  rag(options) {
+    var _a;
+    this.config.rag = __spreadValues(__spreadValues({}, (_a = this.config.rag) != null ? _a : {}), options);
+    return this;
+  }
+  /**
+   * Set UI/UX parameters
+   */
+  ui(options) {
+    this.config.ui = options;
+    return this;
+  }
+  /**
+   * Build and return the configuration
+   */
+  build() {
+    const finalConfig = this.config;
+    return finalConfig;
+  }
+  /**
+   * Build and return as JSON for serialization
+   */
+  toJSON() {
+    return JSON.stringify(this.build());
+  }
+  // ============================================================================
+  // Private helper methods for auto-detection
+  // ============================================================================
+  autoDetectVectorDb() {
+    if (process.env.PINECONE_API_KEY && process.env.PINECONE_INDEX) {
+      return {
+        provider: "pinecone",
+        indexName: process.env.PINECONE_INDEX,
+        options: { apiKey: process.env.PINECONE_API_KEY }
+      };
+    }
+    if (process.env.QDRANT_URL) {
+      return {
+        provider: "qdrant",
+        indexName: process.env.QDRANT_COLLECTION || "documents",
+        options: {
+          url: process.env.QDRANT_URL,
+          apiKey: process.env.QDRANT_API_KEY
+        }
+      };
+    }
+    if (process.env.DATABASE_URL) {
+      return {
+        provider: "postgresql",
+        indexName: process.env.PG_TABLE || "documents",
+        options: {
+          connectionString: process.env.DATABASE_URL
+        }
+      };
+    }
+    if (process.env.MONGODB_URI) {
+      return {
+        provider: "mongodb",
+        indexName: process.env.MONGODB_COLLECTION || "documents",
+        options: {
+          uri: process.env.MONGODB_URI,
+          database: process.env.MONGODB_DB || "ai_db",
+          collection: process.env.MONGODB_COLLECTION || "documents"
+        }
+      };
+    }
+    if (process.env.REDIS_URL) {
+      return {
+        provider: "redis",
+        indexName: process.env.REDIS_INDEX || "documents",
+        options: {
+          url: process.env.REDIS_URL
+        }
+      };
+    }
+    return {
+      provider: "qdrant",
+      indexName: "documents",
+      options: {
+        url: process.env.QDRANT_URL || "http://localhost:6333"
+      }
+    };
+  }
+  autoDetectLLM() {
+    if (process.env.OPENAI_API_KEY) {
+      return {
+        provider: "openai",
+        model: process.env.OPENAI_MODEL || "gpt-4o-mini",
+        apiKey: process.env.OPENAI_API_KEY,
+        maxTokens: parseInt(process.env.OPENAI_MAX_TOKENS || "1024"),
+        temperature: parseFloat(process.env.OPENAI_TEMPERATURE || "0.7")
+      };
+    }
+    if (process.env.ANTHROPIC_API_KEY) {
+      return {
+        provider: "anthropic",
+        model: process.env.ANTHROPIC_MODEL || "claude-3-haiku-20240307",
+        apiKey: process.env.ANTHROPIC_API_KEY,
+        maxTokens: parseInt(process.env.ANTHROPIC_MAX_TOKENS || "1024"),
+        temperature: parseFloat(process.env.ANTHROPIC_TEMPERATURE || "0.7")
+      };
+    }
+    if (process.env.OLLAMA_BASE_URL) {
+      return {
+        provider: "ollama",
+        model: process.env.OLLAMA_MODEL || "mistral",
+        baseUrl: process.env.OLLAMA_BASE_URL,
+        maxTokens: parseInt(process.env.OLLAMA_MAX_TOKENS || "1024"),
+        temperature: parseFloat(process.env.OLLAMA_TEMPERATURE || "0.7")
+      };
+    }
+    return {
+      provider: "openai",
+      model: "gpt-4o-mini",
+      apiKey: process.env.OPENAI_API_KEY
+    };
+  }
+  autoDetectEmbedding() {
+    if (process.env.EMBEDDING_PROVIDER) {
+      const provider = process.env.EMBEDDING_PROVIDER;
+      return {
+        provider,
+        model: process.env.EMBEDDING_MODEL || "default",
+        apiKey: process.env.EMBEDDING_API_KEY,
+        baseUrl: process.env.EMBEDDING_BASE_URL
+      };
+    }
+    return {
+      provider: "openai",
+      model: "text-embedding-3-small",
+      apiKey: process.env.OPENAI_API_KEY
+    };
+  }
+};
+var PRESETS = {
+  /**
+   * OpenAI + Pinecone: Production-ready cloud setup
+   */
+  "openai-pinecone": {
+    vectorDb: "pinecone",
+    llm: "openai",
+    embedding: "openai"
+  },
+  /**
+   * Claude + Qdrant: Open-source vector DB + proprietary LLM
+   */
+  "claude-qdrant": {
+    vectorDb: "qdrant",
+    llm: "anthropic",
+    embedding: "openai"
+  },
+  /**
+   * Local development: Ollama + local Qdrant
+   */
+  "local-dev": {
+    vectorDb: "qdrant",
+    llm: "ollama",
+    embedding: "ollama"
+  },
+  /**
+   * Fully open-source: Ollama LLM + Qdrant vector DB + Ollama embeddings
+   */
+  "fully-open-source": {
+    vectorDb: "qdrant",
+    llm: "ollama",
+    embedding: "ollama"
+  },
+  /**
+   * PostgreSQL stack: pgvector + OpenAI
+   */
+  "postgres-openai": {
+    vectorDb: "postgresql",
+    llm: "openai",
+    embedding: "openai"
+  },
+  /**
+   * Enterprise MongoDB: MongoDB Atlas with OpenAI
+   */
+  "mongodb-openai": {
+    vectorDb: "mongodb",
+    llm: "openai",
+    embedding: "openai"
+  },
+  /**
+   * Redis stack for caching + search
+   */
+  "redis-openai": {
+    vectorDb: "redis",
+    llm: "openai",
+    embedding: "openai"
+  }
+};
+function createFromPreset(presetName) {
+  const preset = PRESETS[presetName];
+  return new ConfigBuilder().vectorDb(preset.vectorDb, { apiKey: process.env[`${preset.vectorDb.toUpperCase()}_API_KEY`] }).llm(preset.llm, void 0, process.env[`${preset.llm.toUpperCase()}_API_KEY`]).embedding(preset.embedding, void 0, process.env[`${preset.embedding.toUpperCase()}_API_KEY`]);
+}
+
+// src/providers/vectordb/MultiTablePostgresProvider.ts
+import { Pool } from "pg";
+var MultiTablePostgresProvider = class extends BaseVectorProvider {
+  constructor(config) {
+    var _a, _b, _c;
+    super(config);
+    const opts = config.options || {};
+    if (!opts.connectionString) {
+      throw new Error("[MultiTablePostgresProvider] options.connectionString is required");
+    }
+    this.connectionString = opts.connectionString;
+    this.dimensions = (_a = opts.dimensions) != null ? _a : 768;
+    const rawTables = (_c = (_b = opts.tables) != null ? _b : process.env.VECTOR_DB_TABLES) != null ? _c : "";
+    this.tables = typeof rawTables === "string" ? rawTables.split(",").map((t) => t.trim()).filter(Boolean) : rawTables;
+    if (this.tables.length === 0) {
+      console.warn(
+        "[MultiTablePostgresProvider] No tables configured. Set VECTOR_DB_TABLES as a comma-separated list of table names or pass options.tables."
+      );
+    }
+  }
+  async initialize() {
+    this.pool = new Pool({ connectionString: this.connectionString });
+    const client = await this.pool.connect();
+    client.release();
+    console.log(
+      `[MultiTablePostgresProvider] Connected. Searching across ${this.tables.length} table(s): ${this.tables.join(", ")}`
+    );
+  }
+  /**
+   * Upsert is not supported for MultiTablePostgresProvider as it's designed for 
+   * searching across pre-existing tables with varying schemas.
+   */
+  async upsert(_doc, _namespace) {
+    throw new Error(
+      "[MultiTablePostgresProvider] upsert() is not supported in multi-table mode. Please use the standard PostgreSQLProvider for single-table managed indices."
+    );
+  }
+  /**
+   * Batch upsert is not supported for MultiTablePostgresProvider.
+   */
+  async batchUpsert(_docs, _namespace) {
+    throw new Error(
+      "[MultiTablePostgresProvider] batchUpsert() is not supported in multi-table mode."
+    );
+  }
+  /**
+   * Query all configured tables and merge results, sorted by cosine similarity score.
+   */
+  async query(vector, topK, _namespace, _filter) {
+    var _b, _c;
+    if (!this.pool) {
+      throw new Error("[MultiTablePostgresProvider] Provider not initialized. Call initialize() first.");
+    }
+    console.log(`[MultiTablePostgresProvider] Querying ${this.tables.length} table(s) with vector dim=${vector.length}`);
+    const vectorLiteral = `[${vector.join(",")}]`;
+    const allResults = [];
+    for (const table of this.tables) {
+      try {
+        const result = await this.pool.query(
+          `SELECT *, 1 - (embedding <=> $1::vector) AS score
+           FROM "${table}"
+           ORDER BY embedding <=> $1::vector
+           LIMIT $2`,
+          [vectorLiteral, topK]
+        );
+        for (const row of result.rows) {
+          const _a = row, { score, id } = _a, rest = __objRest(_a, ["score", "id"]);
+          delete rest.embedding;
+          const content = Object.entries(rest).filter(([k, v]) => v !== null && typeof v !== "object" && k !== "id").map(([k, v]) => `${k}: ${v}`).join("\n");
+          allResults.push({
+            id: `${table}-${id}`,
+            score: parseFloat(String(score)),
+            content,
+            metadata: __spreadProps(__spreadValues({}, rest), { source_table: table })
+          });
+        }
+      } catch (err) {
+        console.error(
+          `[MultiTablePostgresProvider] Error querying table "${table}":`,
+          err.message
+        );
+      }
+    }
+    const sortedResults = allResults.sort((a, b) => b.score - a.score);
+    if (sortedResults.length > 0) {
+      console.log(`[MultiTablePostgresProvider] Top match from "${(_c = (_b = sortedResults[0].metadata) == null ? void 0 : _b.source_table) != null ? _c : "unknown"}" with score ${sortedResults[0].score.toFixed(4)}`);
+    }
+    return sortedResults.slice(0, topK);
+  }
+  async delete(_id, _namespace) {
+    console.warn("[MultiTablePostgresProvider] delete() is a no-op for multi-table mode.");
+  }
+  async deleteNamespace(_namespace) {
+    console.warn("[MultiTablePostgresProvider] deleteNamespace() is a no-op for multi-table mode.");
+  }
+  async ping() {
+    try {
+      await this.pool.query("SELECT 1");
+      return true;
+    } catch (e) {
+      return false;
+    }
+  }
+  async disconnect() {
+    if (this.pool) {
+      await this.pool.end();
+    }
+  }
+};
 export {
   AnthropicProvider,
   BaseVectorProvider,
+  BatchProcessor,
   ChromaDBProvider,
+  ConfigBuilder,
   ConfigResolver,
+  ConfigValidator,
   DocumentChunker,
+  DocumentParser,
+  EmbeddingStrategy,
+  EmbeddingStrategyResolver,
   LLMFactory,
+  LLM_PROFILES,
   MilvusProvider,
   MongoDBProvider,
+  MultiTablePostgresProvider,
   OllamaProvider,
   OpenAIProvider,
+  PRESETS,
   PineconeProvider,
   Pipeline,
   PostgreSQLProvider,
+  ProviderHealthCheck,
   ProviderRegistry,
   QdrantProvider,
   RedisProvider,
+  UniversalLLMAdapter,
+  UniversalVectorProvider,
+  VECTOR_PROFILES,
   VectorPlugin,
   WeaviateProvider,
   createChatHandler,
+  createFromPreset,
   createHealthHandler,
   createIngestHandler,
   createUploadHandler,

package/package.json

@@ -1,22 +1,31 @@
 {
   "name": "@retrivora-ai/rag-engine",
-  "version": "0.1.7",
+  "version": "0.1.9",
   "description": "Retrivora AI is a plug-and-play AI engine for RAG chat experiences — generic vector DB + LLM provider, embeddable or standalone.",
   "author": "Abhinav Alkuchi",
   "license": "MIT",
   "keywords": [
     "rag",
+    "retrieval-augmented-generation",
     "chatbot",
     "vector-database",
     "pinecone",
     "pgvector",
+    "postgresql",
+    "milvus",
+    "qdrant",
+    "chromadb",
+    "redis",
+    "weaviate",
+    "mongodb",
     "openai",
     "anthropic",
     "ollama",
     "nextjs",
     "react",
     "ai",
-    "llm"
+    "llm",
+    "embeddings"
   ],
   "main": "dist/index.js",
   "module": "dist/index.mjs",