@juspay/neurolink/dist/rag/chunkers/MarkdownChunker.d.ts

Universal AI Development Platform with working MCP integration, multi-provider support, voice (TTS/STT/realtime), and professional CLI. 58+ external MCP servers discoverable, multimodal file processing, RAG pipelines. Build, test, and deploy AI applicatio

/**
 * Markdown Chunker
 *
 * Splits markdown content by headers and structural elements.
 * Preserves markdown tables by detecting table boundaries and splitting
 * on row boundaries when a table exceeds the max chunk size.
 */
import type { Chunk, ChunkerConfig, ChunkingStrategy } from "../../types/index.js";
import { BaseChunker } from "./BaseChunker.js";
/**
 * Markdown Chunker
 */
export declare class MarkdownChunker extends BaseChunker {
    readonly strategy: ChunkingStrategy;
    getDefaultConfig(): ChunkerConfig;
    protected doChunk(content: string, config: ChunkerConfig): Promise<Chunk[]>;
    /**
     * Split content while preserving markdown tables.
     *
     * Strategy:
     * 1. Identify table blocks in the content.
     * 2. Split content into segments: non-table text and table blocks.
     * 3. Non-table text is split using paragraph/sentence boundaries (existing logic).
     * 4. Tables that fit in a chunk are kept intact.
     * 5. Oversized tables are split on row boundaries, repeating the header row.
     */
    private splitContentTableAware;
    /**
     * Split a table on row boundaries, repeating header + separator in each chunk.
     */
    private splitTableByRows;
    /**
     * Split non-table text using paragraph and sentence boundaries.
     * This is the original splitContent logic extracted for reuse.
     */
    private splitPlainContent;
}