@retrivora-ai/rag-engine 1.0.4 → 1.0.6

package/dist/DocumentChunker-Dh9TvmGG.d.mts

@@ -0,0 +1,45 @@
+/**
+ * DocumentChunker — splits long text into overlapping chunks
+ * suitable for vector embedding and retrieval.
+ *
+ * Strategy: sentence-aware sliding window
+ *   1. Split on sentence boundaries
+ *   2. Accumulate sentences into chunks up to `chunkSize` characters
+ *   3. Carry over `chunkOverlap` characters from the previous chunk
+ */
+interface Chunk {
+    id: string;
+    content: string;
+    metadata?: Record<string, unknown>;
+}
+interface ChunkOptions {
+    /** Target chunk size in characters (default 1000) */
+    chunkSize?: number;
+    /** Overlap between consecutive chunks in characters (default 200) */
+    chunkOverlap?: number;
+    /** Source document identifier used as ID prefix */
+    docId?: string | number;
+    /** Extra metadata to attach to every chunk */
+    metadata?: Record<string, unknown>;
+    /** Characters used to split text, in order of priority */
+    separators?: string[];
+}
+declare class DocumentChunker {
+    private readonly chunkSize;
+    private readonly chunkOverlap;
+    private readonly separators;
+    constructor(chunkSize?: number, chunkOverlap?: number, separators?: string[]);
+    /**
+     * Split a single text string into overlapping chunks using a recursive strategy.
+     * Preserves structural boundaries (Markdown headers) where possible.
+     */
+    chunk(text: string, options?: ChunkOptions): Chunk[];
+    private recursiveSplit;
+    chunkMany(documents: Array<{
+        content: string;
+        docId?: string | number;
+        metadata?: Record<string, unknown>;
+    }>): Chunk[];
+}
+
+export { type Chunk as C, DocumentChunker as D, type ChunkOptions as a };

package/dist/DocumentChunker-Dh9TvmGG.d.ts

@@ -0,0 +1,45 @@
+/**
+ * DocumentChunker — splits long text into overlapping chunks
+ * suitable for vector embedding and retrieval.
+ *
+ * Strategy: sentence-aware sliding window
+ *   1. Split on sentence boundaries
+ *   2. Accumulate sentences into chunks up to `chunkSize` characters
+ *   3. Carry over `chunkOverlap` characters from the previous chunk
+ */
+interface Chunk {
+    id: string;
+    content: string;
+    metadata?: Record<string, unknown>;
+}
+interface ChunkOptions {
+    /** Target chunk size in characters (default 1000) */
+    chunkSize?: number;
+    /** Overlap between consecutive chunks in characters (default 200) */
+    chunkOverlap?: number;
+    /** Source document identifier used as ID prefix */
+    docId?: string | number;
+    /** Extra metadata to attach to every chunk */
+    metadata?: Record<string, unknown>;
+    /** Characters used to split text, in order of priority */
+    separators?: string[];
+}
+declare class DocumentChunker {
+    private readonly chunkSize;
+    private readonly chunkOverlap;
+    private readonly separators;
+    constructor(chunkSize?: number, chunkOverlap?: number, separators?: string[]);
+    /**
+     * Split a single text string into overlapping chunks using a recursive strategy.
+     * Preserves structural boundaries (Markdown headers) where possible.
+     */
+    chunk(text: string, options?: ChunkOptions): Chunk[];
+    private recursiveSplit;
+    chunkMany(documents: Array<{
+        content: string;
+        docId?: string | number;
+        metadata?: Record<string, unknown>;
+    }>): Chunk[];
+}
+
+export { type Chunk as C, DocumentChunker as D, type ChunkOptions as a };

package/dist/{RagConfig-DRJO4hGU.d.mts → RagConfig-BjC6zSTV.d.mts}

@@ -1,3 +1,60 @@
+/**
+ * Generic LLM Provider interface.
+ * Covers both chat completion and embedding generation so a single
+ * provider can handle both responsibilities when appropriate.
+ */
+interface ChatMessage {
+    role: 'user' | 'assistant' | 'system';
+    content: string;
+}
+interface ChatOptions {
+    /** Override max tokens for this specific call */
+    maxTokens?: number;
+    /** Override temperature for this specific call */
+    temperature?: number;
+    /** Stop sequences */
+    stop?: string[];
+}
+interface EmbedOptions {
+    /** Override model for this specific embed call */
+    model?: string;
+    /** Specify the task type for models that require prefixes (e.g. nomic-embed-text) */
+    taskType?: 'query' | 'document';
+}
+interface ILLMProvider {
+    /**
+     * Send a chat completion request.
+     * @param messages – the full conversation history
+     * @param context  – retrieved RAG context to inject
+     * @param options  – optional per-call overrides
+     * @returns        – the assistant's reply text
+     */
+    chat(messages: ChatMessage[], context: string, options?: ChatOptions): Promise<string>;
+    /**
+     * Send a streaming chat completion request.
+     * @returns – an async iterable of text chunks
+     */
+    chatStream?(messages: ChatMessage[], context: string, options?: ChatOptions): AsyncIterable<string>;
+    /**
+     * Generate an embedding vector for the given text.
+     * @param text    – text to embed
+     * @param options – optional overrides
+     * @returns       – float array (the embedding)
+     */
+    embed(text: string, options?: EmbedOptions): Promise<number[]>;
+    /**
+     * Generate embedding vectors for multiple texts in a single batch.
+     * @param texts   – array of texts to embed
+     * @param options – optional overrides
+     * @returns       – array of float arrays
+     */
+    batchEmbed(texts: string[], options?: EmbedOptions): Promise<number[][]>;
+    /**
+     * Check if the provider endpoint is reachable.
+     */
+    ping(): Promise<boolean>;
+}
+
 interface VectorMatch {
     id: string | number;
     score: number;
@@ -187,6 +244,10 @@ interface RAGConfig {
     useGraphRetrieval?: boolean;
     /** Whether to perform reranking on retrieved results */
     useReranking?: boolean;
+    /** List of metadata fields that are valid for filtering.
+     * Used to dynamically extract hints from natural language queries.
+     */
+    filterableFields?: string[];
 }
 interface RagConfig {
     /**
@@ -208,4 +269,4 @@ interface RagConfig {
     graphDb?: GraphDBConfig;
 }
 
-export type { ChatResponse as C, EmbeddingConfig as E, GraphDBConfig as G, IngestDocument as I, LLMConfig as L, RAGConfig as R, UIConfig as U, VectorMatch as V, EmbeddingProvider as a, LLMProvider as b, RagConfig as c, UpsertDocument as d, VectorDBConfig as e, VectorDBProvider as f, RetrievalResult as g, GraphNode as h, Edge as i, GraphSearchResult as j };
+export type { ChatMessage as C, EmbedOptions as E, GraphDBConfig as G, ILLMProvider as I, LLMConfig as L, RAGConfig as R, UIConfig as U, VectorMatch as V, ChatOptions as a, ChatResponse as b, EmbeddingConfig as c, EmbeddingProvider as d, IngestDocument as e, LLMProvider as f, RagConfig as g, UpsertDocument as h, VectorDBConfig as i, VectorDBProvider as j, RetrievalResult as k, GraphNode as l, Edge as m, GraphSearchResult as n };

package/dist/{RagConfig-DRJO4hGU.d.ts → RagConfig-BjC6zSTV.d.ts}

@@ -1,3 +1,60 @@
+/**
+ * Generic LLM Provider interface.
+ * Covers both chat completion and embedding generation so a single
+ * provider can handle both responsibilities when appropriate.
+ */
+interface ChatMessage {
+    role: 'user' | 'assistant' | 'system';
+    content: string;
+}
+interface ChatOptions {
+    /** Override max tokens for this specific call */
+    maxTokens?: number;
+    /** Override temperature for this specific call */
+    temperature?: number;
+    /** Stop sequences */
+    stop?: string[];
+}
+interface EmbedOptions {
+    /** Override model for this specific embed call */
+    model?: string;
+    /** Specify the task type for models that require prefixes (e.g. nomic-embed-text) */
+    taskType?: 'query' | 'document';
+}
+interface ILLMProvider {
+    /**
+     * Send a chat completion request.
+     * @param messages – the full conversation history
+     * @param context  – retrieved RAG context to inject
+     * @param options  – optional per-call overrides
+     * @returns        – the assistant's reply text
+     */
+    chat(messages: ChatMessage[], context: string, options?: ChatOptions): Promise<string>;
+    /**
+     * Send a streaming chat completion request.
+     * @returns – an async iterable of text chunks
+     */
+    chatStream?(messages: ChatMessage[], context: string, options?: ChatOptions): AsyncIterable<string>;
+    /**
+     * Generate an embedding vector for the given text.
+     * @param text    – text to embed
+     * @param options – optional overrides
+     * @returns       – float array (the embedding)
+     */
+    embed(text: string, options?: EmbedOptions): Promise<number[]>;
+    /**
+     * Generate embedding vectors for multiple texts in a single batch.
+     * @param texts   – array of texts to embed
+     * @param options – optional overrides
+     * @returns       – array of float arrays
+     */
+    batchEmbed(texts: string[], options?: EmbedOptions): Promise<number[][]>;
+    /**
+     * Check if the provider endpoint is reachable.
+     */
+    ping(): Promise<boolean>;
+}
+
 interface VectorMatch {
     id: string | number;
     score: number;
@@ -187,6 +244,10 @@ interface RAGConfig {
     useGraphRetrieval?: boolean;
     /** Whether to perform reranking on retrieved results */
     useReranking?: boolean;
+    /** List of metadata fields that are valid for filtering.
+     * Used to dynamically extract hints from natural language queries.
+     */
+    filterableFields?: string[];
 }
 interface RagConfig {
     /**
@@ -208,4 +269,4 @@ interface RagConfig {
     graphDb?: GraphDBConfig;
 }
 
-export type { ChatResponse as C, EmbeddingConfig as E, GraphDBConfig as G, IngestDocument as I, LLMConfig as L, RAGConfig as R, UIConfig as U, VectorMatch as V, EmbeddingProvider as a, LLMProvider as b, RagConfig as c, UpsertDocument as d, VectorDBConfig as e, VectorDBProvider as f, RetrievalResult as g, GraphNode as h, Edge as i, GraphSearchResult as j };
+export type { ChatMessage as C, EmbedOptions as E, GraphDBConfig as G, ILLMProvider as I, LLMConfig as L, RAGConfig as R, UIConfig as U, VectorMatch as V, ChatOptions as a, ChatResponse as b, EmbeddingConfig as c, EmbeddingProvider as d, IngestDocument as e, LLMProvider as f, RagConfig as g, UpsertDocument as h, VectorDBConfig as i, VectorDBProvider as j, RetrievalResult as k, GraphNode as l, Edge as m, GraphSearchResult as n };