Public declaration syntax from forge-ts/packages/forge-embed/src/chunking.ts
Original source SHA-256: c78ef30a78111bcfeb5c5ed82e48a59c3144035d43a0961485dcaded4acb2e93
Function bodies and constant values are omitted. This is not the complete implementation.

Source line 26
export interface TextSplitter {
  /**
   * Splits text into chunks.
   *
   * @param text - The text to split.
   * @returns An array of text chunks.
   */
  split(text: string): string[];
}

Source line 39
export interface RecursiveCharacterSplitterOptions {
  /** Maximum number of characters per chunk. Defaults to 1000. */
  readonly chunkSize?: number;
  /** Number of overlapping characters between consecutive chunks. Defaults to 200. */
  readonly chunkOverlap?: number;
  /**
   * Separator hierarchy from coarsest to finest. Defaults to
   * `['\n\n', '\n', '. ', ' ', '']` (paragraphs, lines, sentences, words, chars).
   */
  readonly separators?: readonly string[];
}

Source line 70
export class RecursiveCharacterSplitter implements TextSplitter {
  constructor(options: RecursiveCharacterSplitterOptions = {});
  split(text: string): string[];
}

Source line 232
export interface TokenSplitterOptions {
  /** Maximum number of tokens per chunk. Defaults to 256. */
  readonly tokensPerChunk?: number;
  /** Number of overlapping tokens between consecutive chunks. Defaults to 0. */
  readonly tokenOverlap?: number;
}

Source line 252
export class TokenSplitter implements TextSplitter {
  constructor(options: TokenSplitterOptions = {});
  split(text: string): string[];
}