@superatomai/sdk-node 0.0.24 → 0.0.25

package/dist/index.d.mts

@@ -1483,6 +1483,134 @@ declare const CONTEXT_CONFIG: {
     MAX_CONVERSATION_CONTEXT_BLOCKS: number;
 };
 
+/**
+ * BM25L Reranker for hybrid semantic search
+ *
+ * BM25L is an improved variant of BM25 that provides better handling of
+ * long documents and term frequency saturation. This implementation is
+ * designed to rerank semantic search results from ChromaDB.
+ *
+ * The hybrid approach combines:
+ * 1. Semantic similarity from ChromaDB embeddings (dense vectors)
+ * 2. Lexical matching from BM25L (sparse, keyword-based)
+ *
+ * This addresses the weakness of pure semantic search which may miss
+ * exact keyword matches that are important for user intent.
+ */
+interface BM25LOptions {
+    /** Term frequency saturation parameter (default: 1.5) */
+    k1?: number;
+    /** Length normalization parameter (default: 0.75) */
+    b?: number;
+    /** Lower-bound adjustment from BM25L paper (default: 0.5) */
+    delta?: number;
+}
+interface RerankedResult<T> {
+    item: T;
+    originalIndex: number;
+    semanticScore: number;
+    bm25Score: number;
+    hybridScore: number;
+}
+interface HybridSearchOptions extends BM25LOptions {
+    /** Weight for semantic score (0-1, default: 0.7) */
+    semanticWeight?: number;
+    /** Weight for BM25 score (0-1, default: 0.3) */
+    bm25Weight?: number;
+    /** Minimum hybrid score threshold (0-1, default: 0) */
+    minScore?: number;
+}
+/**
+ * BM25L implementation for lexical scoring
+ */
+declare class BM25L {
+    private k1;
+    private b;
+    private delta;
+    private documents;
+    private docLengths;
+    private avgDocLength;
+    private termDocFreq;
+    /**
+     * @param documents - Array of raw documents (strings)
+     * @param opts - Optional BM25L parameters
+     */
+    constructor(documents?: string[], opts?: BM25LOptions);
+    /**
+     * Tokenize text into lowercase alphanumeric tokens
+     */
+    tokenize(text: string): string[];
+    /**
+     * Compute IDF (Inverse Document Frequency) with smoothing
+     */
+    private idf;
+    /**
+     * Compute BM25L score for a single document
+     */
+    score(query: string, docIndex: number): number;
+    /**
+     * Search and rank all documents
+     */
+    search(query: string): Array<{
+        index: number;
+        score: number;
+    }>;
+}
+/**
+ * Hybrid reranker that combines semantic and BM25L scores
+ *
+ * @param query - The search query
+ * @param items - Array of items to rerank
+ * @param getDocument - Function to extract document text from an item
+ * @param getSemanticScore - Function to extract semantic similarity score from an item
+ * @param options - Hybrid search options
+ * @returns Reranked items with hybrid scores
+ */
+declare function hybridRerank<T>(query: string, items: T[], getDocument: (item: T) => string, getSemanticScore: (item: T) => number, options?: HybridSearchOptions): RerankedResult<T>[];
+/**
+ * Simple reranking function for ChromaDB results
+ *
+ * This is a convenience wrapper for reranking ChromaDB query results
+ * that follow the standard { ids, documents, metadatas, distances } format.
+ *
+ * @param query - The search query
+ * @param chromaResults - ChromaDB query results
+ * @param options - Hybrid search options
+ * @returns Reranked results with hybrid scores
+ */
+declare function rerankChromaResults(query: string, chromaResults: {
+    ids: string[][];
+    documents: (string | null)[][];
+    metadatas: Record<string, any>[][];
+    distances: number[][];
+}, options?: HybridSearchOptions): Array<{
+    id: string;
+    document: string | null;
+    metadata: Record<string, any>;
+    distance: number;
+    semanticScore: number;
+    bm25Score: number;
+    hybridScore: number;
+}>;
+/**
+ * Rerank conversation search results specifically
+ *
+ * This function is designed to work with the conversation-history.search collection
+ * where we need to fetch more results initially and then rerank them.
+ *
+ * @param query - The user's search query
+ * @param results - Array of conversation search results from ChromaDB
+ * @param options - Hybrid search options
+ * @returns Reranked results sorted by hybrid score
+ */
+declare function rerankConversationResults<T extends {
+    userPrompt?: string;
+    similarity?: number;
+}>(query: string, results: T[], options?: HybridSearchOptions): Array<T & {
+    hybridScore: number;
+    bm25Score: number;
+}>;
+
 declare const SDK_VERSION = "0.0.8";
 type MessageTypeHandler = (message: IncomingMessage) => void | Promise<void>;
 declare class SuperatomSDK {
@@ -1587,4 +1715,4 @@ declare class SuperatomSDK {
     getTools(): Tool$1[];
 }
 
-export { type Action, CONTEXT_CONFIG, type CapturedLog, CleanupService, type CollectionHandler, type CollectionOperation, type DBUIBlock, type IncomingMessage, type KbNodesQueryFilters, type KbNodesRequestPayload, LLM, type LogLevel, type Message, SDK_VERSION, STORAGE_CONFIG, SuperatomSDK, type SuperatomSDKConfig, Thread, ThreadManager, type Tool$1 as Tool, UIBlock, UILogCollector, type User, UserManager, type UsersData, logger };
+export { type Action, BM25L, type BM25LOptions, CONTEXT_CONFIG, type CapturedLog, CleanupService, type CollectionHandler, type CollectionOperation, type DBUIBlock, type HybridSearchOptions, type IncomingMessage, type KbNodesQueryFilters, type KbNodesRequestPayload, LLM, type LogLevel, type Message, type RerankedResult, SDK_VERSION, STORAGE_CONFIG, SuperatomSDK, type SuperatomSDKConfig, Thread, ThreadManager, type Tool$1 as Tool, UIBlock, UILogCollector, type User, UserManager, type UsersData, hybridRerank, logger, rerankChromaResults, rerankConversationResults };

package/dist/index.d.ts

@@ -1483,6 +1483,134 @@ declare const CONTEXT_CONFIG: {
     MAX_CONVERSATION_CONTEXT_BLOCKS: number;
 };
 
+/**
+ * BM25L Reranker for hybrid semantic search
+ *
+ * BM25L is an improved variant of BM25 that provides better handling of
+ * long documents and term frequency saturation. This implementation is
+ * designed to rerank semantic search results from ChromaDB.
+ *
+ * The hybrid approach combines:
+ * 1. Semantic similarity from ChromaDB embeddings (dense vectors)
+ * 2. Lexical matching from BM25L (sparse, keyword-based)
+ *
+ * This addresses the weakness of pure semantic search which may miss
+ * exact keyword matches that are important for user intent.
+ */
+interface BM25LOptions {
+    /** Term frequency saturation parameter (default: 1.5) */
+    k1?: number;
+    /** Length normalization parameter (default: 0.75) */
+    b?: number;
+    /** Lower-bound adjustment from BM25L paper (default: 0.5) */
+    delta?: number;
+}
+interface RerankedResult<T> {
+    item: T;
+    originalIndex: number;
+    semanticScore: number;
+    bm25Score: number;
+    hybridScore: number;
+}
+interface HybridSearchOptions extends BM25LOptions {
+    /** Weight for semantic score (0-1, default: 0.7) */
+    semanticWeight?: number;
+    /** Weight for BM25 score (0-1, default: 0.3) */
+    bm25Weight?: number;
+    /** Minimum hybrid score threshold (0-1, default: 0) */
+    minScore?: number;
+}
+/**
+ * BM25L implementation for lexical scoring
+ */
+declare class BM25L {
+    private k1;
+    private b;
+    private delta;
+    private documents;
+    private docLengths;
+    private avgDocLength;
+    private termDocFreq;
+    /**
+     * @param documents - Array of raw documents (strings)
+     * @param opts - Optional BM25L parameters
+     */
+    constructor(documents?: string[], opts?: BM25LOptions);
+    /**
+     * Tokenize text into lowercase alphanumeric tokens
+     */
+    tokenize(text: string): string[];
+    /**
+     * Compute IDF (Inverse Document Frequency) with smoothing
+     */
+    private idf;
+    /**
+     * Compute BM25L score for a single document
+     */
+    score(query: string, docIndex: number): number;
+    /**
+     * Search and rank all documents
+     */
+    search(query: string): Array<{
+        index: number;
+        score: number;
+    }>;
+}
+/**
+ * Hybrid reranker that combines semantic and BM25L scores
+ *
+ * @param query - The search query
+ * @param items - Array of items to rerank
+ * @param getDocument - Function to extract document text from an item
+ * @param getSemanticScore - Function to extract semantic similarity score from an item
+ * @param options - Hybrid search options
+ * @returns Reranked items with hybrid scores
+ */
+declare function hybridRerank<T>(query: string, items: T[], getDocument: (item: T) => string, getSemanticScore: (item: T) => number, options?: HybridSearchOptions): RerankedResult<T>[];
+/**
+ * Simple reranking function for ChromaDB results
+ *
+ * This is a convenience wrapper for reranking ChromaDB query results
+ * that follow the standard { ids, documents, metadatas, distances } format.
+ *
+ * @param query - The search query
+ * @param chromaResults - ChromaDB query results
+ * @param options - Hybrid search options
+ * @returns Reranked results with hybrid scores
+ */
+declare function rerankChromaResults(query: string, chromaResults: {
+    ids: string[][];
+    documents: (string | null)[][];
+    metadatas: Record<string, any>[][];
+    distances: number[][];
+}, options?: HybridSearchOptions): Array<{
+    id: string;
+    document: string | null;
+    metadata: Record<string, any>;
+    distance: number;
+    semanticScore: number;
+    bm25Score: number;
+    hybridScore: number;
+}>;
+/**
+ * Rerank conversation search results specifically
+ *
+ * This function is designed to work with the conversation-history.search collection
+ * where we need to fetch more results initially and then rerank them.
+ *
+ * @param query - The user's search query
+ * @param results - Array of conversation search results from ChromaDB
+ * @param options - Hybrid search options
+ * @returns Reranked results sorted by hybrid score
+ */
+declare function rerankConversationResults<T extends {
+    userPrompt?: string;
+    similarity?: number;
+}>(query: string, results: T[], options?: HybridSearchOptions): Array<T & {
+    hybridScore: number;
+    bm25Score: number;
+}>;
+
 declare const SDK_VERSION = "0.0.8";
 type MessageTypeHandler = (message: IncomingMessage) => void | Promise<void>;
 declare class SuperatomSDK {
@@ -1587,4 +1715,4 @@ declare class SuperatomSDK {
     getTools(): Tool$1[];
 }
 
-export { type Action, CONTEXT_CONFIG, type CapturedLog, CleanupService, type CollectionHandler, type CollectionOperation, type DBUIBlock, type IncomingMessage, type KbNodesQueryFilters, type KbNodesRequestPayload, LLM, type LogLevel, type Message, SDK_VERSION, STORAGE_CONFIG, SuperatomSDK, type SuperatomSDKConfig, Thread, ThreadManager, type Tool$1 as Tool, UIBlock, UILogCollector, type User, UserManager, type UsersData, logger };
+export { type Action, BM25L, type BM25LOptions, CONTEXT_CONFIG, type CapturedLog, CleanupService, type CollectionHandler, type CollectionOperation, type DBUIBlock, type HybridSearchOptions, type IncomingMessage, type KbNodesQueryFilters, type KbNodesRequestPayload, LLM, type LogLevel, type Message, type RerankedResult, SDK_VERSION, STORAGE_CONFIG, SuperatomSDK, type SuperatomSDKConfig, Thread, ThreadManager, type Tool$1 as Tool, UIBlock, UILogCollector, type User, UserManager, type UsersData, hybridRerank, logger, rerankChromaResults, rerankConversationResults };