@kreuzberg/node 4.0.0-rc.5

package/index.d.ts

@@ -0,0 +1,1109 @@
+/* auto-generated by NAPI-RS */
+/* eslint-disable */
+/**
+ * Batch extract from multiple byte arrays (asynchronous).
+ *
+ * Asynchronously processes multiple in-memory buffers in parallel. Non-blocking
+ * alternative to `batchExtractBytesSync`.
+ *
+ * # Parameters
+ *
+ * * `data_list` - Array of buffers to extract
+ * * `mime_types` - Array of MIME types (must match data_list length)
+ * * `config` - Optional extraction configuration
+ *
+ * # Returns
+ *
+ * Promise resolving to array of `ExtractionResult`.
+ *
+ * # Example
+ *
+ * ```typescript
+ * import { batchExtractBytes } from '@kreuzberg/node';
+ *
+ * const responses = await Promise.all([
+ *   fetch('https://example.com/doc1.pdf'),
+ *   fetch('https://example.com/doc2.pdf')
+ * ]);
+ * const buffers = await Promise.all(
+ *   responses.map(r => r.arrayBuffer().then(b => Buffer.from(b)))
+ * );
+ * const results = await batchExtractBytes(
+ *   buffers,
+ *   ['application/pdf', 'application/pdf'],
+ *   null
+ * );
+ * ```
+ */
+export declare function batchExtractBytes(dataList: Array<Buffer>, mimeTypes: Array<string>, config?: JsExtractionConfig | undefined | null): Promise<Array<JsExtractionResult>>
+
+/**
+ * Batch extract from multiple byte arrays (synchronous).
+ *
+ * Synchronously processes multiple in-memory buffers in parallel. Requires
+ * corresponding MIME types for each buffer.
+ *
+ * # Parameters
+ *
+ * * `data_list` - Array of buffers to extract
+ * * `mime_types` - Array of MIME types (must match data_list length)
+ * * `config` - Optional extraction configuration
+ *
+ * # Returns
+ *
+ * Array of `ExtractionResult` in the same order as inputs.
+ *
+ * # Errors
+ *
+ * Throws if data_list and mime_types lengths don't match.
+ *
+ * # Example
+ *
+ * ```typescript
+ * import { batchExtractBytesSync } from '@kreuzberg/node';
+ *
+ * const buffers = [buffer1, buffer2, buffer3];
+ * const mimeTypes = ['application/pdf', 'image/png', 'text/plain'];
+ * const results = batchExtractBytesSync(buffers, mimeTypes, null);
+ * ```
+ */
+export declare function batchExtractBytesSync(dataList: Array<Buffer>, mimeTypes: Array<string>, config?: JsExtractionConfig | undefined | null): Array<JsExtractionResult>
+
+/**
+ * Batch extract from multiple files (asynchronous).
+ *
+ * Asynchronously processes multiple files in parallel. Non-blocking alternative
+ * to `batchExtractFilesSync` with same performance benefits.
+ *
+ * # Parameters
+ *
+ * * `paths` - Array of file paths to extract
+ * * `config` - Optional extraction configuration (applied to all files)
+ *
+ * # Returns
+ *
+ * Promise resolving to array of `ExtractionResult`.
+ *
+ * # Example
+ *
+ * ```typescript
+ * import { batchExtractFiles } from '@kreuzberg/node';
+ *
+ * const files = ['report1.pdf', 'report2.pdf', 'report3.pdf'];
+ * const results = await batchExtractFiles(files, null);
+ * console.log(`Processed ${results.length} files`);
+ * ```
+ */
+export declare function batchExtractFiles(paths: Array<string>, config?: JsExtractionConfig | undefined | null): Promise<Array<JsExtractionResult>>
+
+/**
+ * Batch extract from multiple files (synchronous).
+ *
+ * Synchronously processes multiple files in parallel using Rayon. Significantly
+ * faster than sequential processing for large batches.
+ *
+ * # Parameters
+ *
+ * * `paths` - Array of file paths to extract
+ * * `config` - Optional extraction configuration (applied to all files)
+ *
+ * # Returns
+ *
+ * Array of `ExtractionResult` in the same order as input paths.
+ *
+ * # Example
+ *
+ * ```typescript
+ * import { batchExtractFilesSync } from '@kreuzberg/node';
+ *
+ * const files = ['doc1.pdf', 'doc2.docx', 'doc3.txt'];
+ * const results = batchExtractFilesSync(files, null);
+ * results.forEach((result, i) => {
+ *   console.log(`File ${files[i]}: ${result.content.substring(0, 100)}...`);
+ * });
+ * ```
+ */
+export declare function batchExtractFilesSync(paths: Array<string>, config?: JsExtractionConfig | undefined | null): Array<JsExtractionResult>
+
+/**
+ * Clear all registered document extractors.
+ *
+ * Removes all document extractors from the registry, including built-in extractors.
+ * Use with caution as this will make document extraction unavailable until
+ * extractors are re-registered.
+ *
+ * # Example
+ *
+ * ```typescript
+ * import { clearDocumentExtractors } from 'kreuzberg';
+ *
+ * clearDocumentExtractors();
+ * ```
+ */
+export declare function clearDocumentExtractors(): void
+
+/**
+ * Clear all registered OCR backends.
+ *
+ * Removes all OCR backends from the registry, including built-in backends.
+ * Use with caution as this will make OCR functionality unavailable until
+ * backends are re-registered.
+ *
+ * # Example
+ *
+ * ```typescript
+ * import { clearOcrBackends } from 'kreuzberg';
+ *
+ * clearOcrBackends();
+ * ```
+ */
+export declare function clearOcrBackends(): void
+
+/** Clear all registered postprocessors */
+export declare function clearPostProcessors(): void
+
+/** Clear all registered validators */
+export declare function clearValidators(): void
+
+/**
+ * Detect MIME type from raw bytes.
+ *
+ * Uses content inspection (magic bytes) to determine MIME type.
+ * This is more accurate than extension-based detection but requires
+ * reading the file content.
+ *
+ * # Parameters
+ *
+ * * `bytes` - Raw file content as Buffer
+ *
+ * # Returns
+ *
+ * The detected MIME type string.
+ *
+ * # Errors
+ *
+ * Throws an error if MIME type cannot be determined from content.
+ *
+ * # Example
+ *
+ * ```typescript
+ * import { detectMimeType } from 'kreuzberg';
+ * import * as fs from 'fs';
+ *
+ * // Read file content
+ * const content = fs.readFileSync('document.pdf');
+ *
+ * // Detect MIME type from bytes
+ * const mimeType = detectMimeType(content);
+ * console.log(mimeType); // 'application/pdf'
+ * ```
+ */
+export declare function detectMimeType(bytes: Buffer): string
+
+/**
+ * Detect MIME type from a file path.
+ *
+ * Uses file extension to determine MIME type. Falls back to `mime_guess` crate
+ * if extension-based detection fails.
+ *
+ * # Parameters
+ *
+ * * `path` - Path to the file (string)
+ * * `check_exists` - Whether to verify file existence (default: true)
+ *
+ * # Returns
+ *
+ * The detected MIME type string.
+ *
+ * # Errors
+ *
+ * Throws an error if:
+ * - File doesn't exist (when check_exists is true)
+ * - MIME type cannot be determined from path/extension
+ * - Extension is unknown
+ *
+ * # Example
+ *
+ * ```typescript
+ * import { detectMimeTypeFromPath } from 'kreuzberg';
+ *
+ * // Detect from existing file
+ * const mimeType = detectMimeTypeFromPath('document.pdf');
+ * console.log(mimeType); // 'application/pdf'
+ *
+ * const mimeType2 = detectMimeTypeFromPath('document.docx');
+ * console.log(mimeType2); // 'application/vnd.openxmlformats-officedocument.wordprocessingml.document'
+ * ```
+ */
+export declare function detectMimeTypeFromPath(path: string, checkExists?: boolean | undefined | null): string
+
+/**
+ * Discover and load extraction configuration from current or parent directories.
+ *
+ * Searches for a `kreuzberg.toml` file starting from the current working directory
+ * and traversing up the directory tree. Returns the first configuration file found.
+ *
+ * # Returns
+ *
+ * `JsExtractionConfig` object if a configuration file is found, or `null` if no
+ * configuration file exists in the current or parent directories.
+ *
+ * # Example
+ *
+ * ```typescript
+ * import { ExtractionConfig } from 'kreuzberg';
+ *
+ * // Try to find config in current or parent directories
+ * const config = ExtractionConfig.discover();
+ * if (config) {
+ *   console.log('Found configuration');
+ *   // Use config for extraction
+ * } else {
+ *   console.log('No configuration file found, using defaults');
+ * }
+ * ```
+ */
+export declare function discoverExtractionConfig(): JsExtractionConfig | null
+
+/**
+ * Embedding preset configuration for TypeScript bindings.
+ *
+ * Contains all settings for a specific embedding model preset.
+ */
+export interface EmbeddingPreset {
+  /** Name of the preset (e.g., "fast", "balanced", "quality", "multilingual") */
+  name: string
+  /** Recommended chunk size in characters */
+  chunkSize: number
+  /** Recommended overlap in characters */
+  overlap: number
+  /** Model identifier (e.g., "AllMiniLML6V2Q", "BGEBaseENV15") */
+  modelName: string
+  /** Embedding vector dimensions */
+  dimensions: number
+  /** Human-readable description of the preset */
+  description: string
+}
+
+/**
+ * Extract content from bytes (asynchronous).
+ *
+ * Asynchronously extracts content from a byte buffer. Non-blocking alternative
+ * to `extractBytesSync` for processing in-memory data.
+ *
+ * # Parameters
+ *
+ * * `data` - Buffer containing the document bytes
+ * * `mime_type` - MIME type of the data
+ * * `config` - Optional extraction configuration
+ *
+ * # Returns
+ *
+ * Promise resolving to `ExtractionResult`.
+ *
+ * # Example
+ *
+ * ```typescript
+ * import { extractBytes } from '@kreuzberg/node';
+ *
+ * const response = await fetch('https://example.com/document.pdf');
+ * const buffer = Buffer.from(await response.arrayBuffer());
+ * const result = await extractBytes(buffer, 'application/pdf', null);
+ * ```
+ */
+export declare function extractBytes(data: Buffer, mimeType: string, config?: JsExtractionConfig | undefined | null): Promise<JsExtractionResult>
+
+/**
+ * Extract content from bytes (synchronous).
+ *
+ * Synchronously extracts content from a byte buffer without requiring a file path.
+ * Useful for processing in-memory data, network streams, or database BLOBs.
+ *
+ * # Parameters
+ *
+ * * `data` - Buffer containing the document bytes
+ * * `mime_type` - MIME type of the data (e.g., "application/pdf", "image/png")
+ * * `config` - Optional extraction configuration
+ *
+ * # Returns
+ *
+ * `ExtractionResult` with extracted content and metadata.
+ *
+ * # Errors
+ *
+ * Throws an error if data is malformed or MIME type is unsupported.
+ *
+ * # Example
+ *
+ * ```typescript
+ * import { extractBytesSync } from '@kreuzberg/node';
+ * import fs from 'fs';
+ *
+ * const buffer = fs.readFileSync('document.pdf');
+ * const result = extractBytesSync(buffer, 'application/pdf', null);
+ * console.log(result.content);
+ * ```
+ */
+export declare function extractBytesSync(data: Buffer, mimeType: string, config?: JsExtractionConfig | undefined | null): JsExtractionResult
+
+/**
+ * Extract content from a file (asynchronous).
+ *
+ * Asynchronously extracts text, tables, images, and metadata from a document file.
+ * Non-blocking alternative to `extractFileSync` for use in async/await contexts.
+ *
+ * # Parameters
+ *
+ * * `file_path` - Path to the file to extract (absolute or relative)
+ * * `mime_type` - Optional MIME type hint (auto-detected if omitted)
+ * * `config` - Optional extraction configuration (OCR, chunking, etc.)
+ *
+ * # Returns
+ *
+ * Promise resolving to `ExtractionResult` with extracted content and metadata.
+ *
+ * # Errors
+ *
+ * Rejects if file processing fails (see `extractFileSync` for error conditions).
+ *
+ * # Example
+ *
+ * ```typescript
+ * import { extractFile } from '@kreuzberg/node';
+ *
+ * // Async/await usage
+ * const result = await extractFile('document.pdf', null, null);
+ * console.log(result.content);
+ *
+ * // Promise usage
+ * extractFile('report.docx', null, null)
+ *   .then(result => console.log(result.content))
+ *   .catch(err => console.error(err));
+ * ```
+ */
+export declare function extractFile(filePath: string, mimeType?: string | undefined | null, config?: JsExtractionConfig | undefined | null): Promise<JsExtractionResult>
+
+/**
+ * Extract content from a file (synchronous).
+ *
+ * Synchronously extracts text, tables, images, and metadata from a document file.
+ * Supports 118+ file formats including PDFs, Office documents, images, and more.
+ *
+ * # Parameters
+ *
+ * * `file_path` - Path to the file to extract (absolute or relative)
+ * * `mime_type` - Optional MIME type hint (auto-detected if omitted)
+ * * `config` - Optional extraction configuration (OCR, chunking, etc.)
+ *
+ * # Returns
+ *
+ * `ExtractionResult` containing:
+ * - `content`: Extracted text content
+ * - `mimeType`: Detected MIME type
+ * - `metadata`: File metadata (author, title, etc.)
+ * - `tables`: Extracted tables (if any)
+ * - `images`: Extracted images (if configured)
+ * - `chunks`: Text chunks (if chunking enabled)
+ * - `detectedLanguages`: Detected languages (if enabled)
+ *
+ * # Errors
+ *
+ * Throws an error if:
+ * - File does not exist or is not accessible
+ * - File format is unsupported
+ * - File is corrupted or malformed
+ * - OCR processing fails (if enabled)
+ *
+ * # Example
+ *
+ * ```typescript
+ * import { extractFileSync, ExtractionConfig } from '@kreuzberg/node';
+ *
+ * // Basic extraction
+ * const result = extractFileSync('document.pdf', null, null);
+ * console.log(result.content);
+ *
+ * // With MIME type hint
+ * const result2 = extractFileSync('file.bin', 'application/pdf', null);
+ *
+ * // With OCR enabled
+ * const config: ExtractionConfig = {
+ *   ocr: {
+ *     backend: 'tesseract',
+ *     language: 'eng',
+ *   }
+ * };
+ * const result3 = extractFileSync('scanned.pdf', null, config);
+ * ```
+ */
+export declare function extractFileSync(filePath: string, mimeType?: string | undefined | null, config?: JsExtractionConfig | undefined | null): JsExtractionResult
+
+/**
+ * Get a specific embedding preset by name.
+ *
+ * Returns a preset configuration object, or null if the preset name is not found.
+ *
+ * # Arguments
+ *
+ * * `name` - The preset name (case-sensitive)
+ *
+ * # Returns
+ *
+ * An `EmbeddingPreset` object with the following properties:
+ * - `name`: string - Preset name
+ * - `chunkSize`: number - Recommended chunk size in characters
+ * - `overlap`: number - Recommended overlap in characters
+ * - `modelName`: string - Model identifier
+ * - `dimensions`: number - Embedding vector dimensions
+ * - `description`: string - Human-readable description
+ *
+ * Returns `null` if preset name is not found.
+ *
+ * # Example
+ *
+ * ```typescript
+ * import { getEmbeddingPreset } from 'kreuzberg';
+ *
+ * const preset = getEmbeddingPreset('balanced');
+ * if (preset) {
+ *   console.log(`Model: ${preset.modelName}, Dims: ${preset.dimensions}`);
+ *   // Model: BGEBaseENV15, Dims: 768
+ * }
+ * ```
+ */
+export declare function getEmbeddingPreset(name: string): EmbeddingPreset | null
+
+/**
+ * Get file extensions for a given MIME type.
+ *
+ * Returns an array of file extensions commonly associated with the specified
+ * MIME type. For example, 'application/pdf' returns ['pdf'].
+ *
+ * # Parameters
+ *
+ * * `mime_type` - The MIME type to look up (e.g., 'application/pdf', 'image/jpeg')
+ *
+ * # Returns
+ *
+ * Array of file extensions (without leading dots).
+ *
+ * # Errors
+ *
+ * Throws an error if the MIME type is not recognized or supported.
+ *
+ * # Example
+ *
+ * ```typescript
+ * import { getExtensionsForMime } from 'kreuzberg';
+ *
+ * // Get extensions for PDF
+ * const pdfExts = getExtensionsForMime('application/pdf');
+ * console.log(pdfExts); // ['pdf']
+ *
+ * // Get extensions for JPEG
+ * const jpegExts = getExtensionsForMime('image/jpeg');
+ * console.log(jpegExts); // ['jpg', 'jpeg']
+ * ```
+ */
+export declare function getExtensionsForMime(mimeType: string): Array<string>
+
+/**
+ * Get the error code for the last FFI error.
+ *
+ * Returns the FFI error code as an integer. Error codes are:
+ * - 0: Success (no error)
+ * - 1: GenericError
+ * - 2: Panic
+ * - 3: InvalidArgument
+ * - 4: IoError
+ * - 5: ParsingError
+ * - 6: OcrError
+ * - 7: MissingDependency
+ *
+ * This is useful for programmatic error handling and distinguishing
+ * between different types of failures in native code.
+ *
+ * # Returns
+ *
+ * The integer error code.
+ *
+ * # Example
+ *
+ * ```typescript
+ * import { extractFile, getLastErrorCode, ErrorCode } from '@kreuzberg/node';
+ *
+ * try {
+ *   const result = await extractFile('document.pdf');
+ * } catch (error) {
+ *   const code = getLastErrorCode();
+ *   if (code === ErrorCode.Panic) {
+ *     console.error('Native code panic detected');
+ *   }
+ * }
+ * ```
+ */
+export declare function getLastErrorCode(): number
+
+/**
+ * Get panic context information if the last error was a panic.
+ *
+ * Returns detailed information about a panic in native code, or null
+ * if the last error was not a panic.
+ *
+ * # Returns
+ *
+ * A `PanicContext` object with:
+ * - `file`: string - Source file where panic occurred
+ * - `line`: number - Line number
+ * - `function`: string - Function name
+ * - `message`: string - Panic message
+ * - `timestamp_secs`: number - Unix timestamp (seconds since epoch)
+ *
+ * Returns `null` if no panic context is available.
+ *
+ * # Example
+ *
+ * ```typescript
+ * import { extractFile, getLastPanicContext } from '@kreuzberg/node';
+ *
+ * try {
+ *   const result = await extractFile('document.pdf');
+ * } catch (error) {
+ *   const context = getLastPanicContext();
+ *   if (context) {
+ *     console.error(`Panic at ${context.file}:${context.line}`);
+ *     console.error(`In function: ${context.function}`);
+ *     console.error(`Message: ${context.message}`);
+ *   }
+ * }
+ * ```
+ */
+export declare function getLastPanicContext(): any | null
+
+export interface JsChunk {
+  content: string
+  embedding?: number[] | undefined
+  metadata: JsChunkMetadata
+}
+
+export interface JsChunkingConfig {
+  maxChars?: number
+  maxOverlap?: number
+  /** Optional embedding configuration for generating embeddings */
+  embedding?: JsEmbeddingConfig
+  /** Optional preset name for chunking parameters */
+  preset?: string
+}
+
+export interface JsChunkMetadata {
+  charStart: number
+  charEnd: number
+  tokenCount?: number
+  chunkIndex: number
+  totalChunks: number
+}
+
+/** Embedding generation configuration for Node.js bindings. */
+export interface JsEmbeddingConfig {
+  /** Embedding model configuration */
+  model?: JsEmbeddingModelType
+  /** Whether to normalize embeddings (L2 normalization) */
+  normalize?: boolean
+  /** Batch size for embedding generation */
+  batchSize?: number
+  /** Whether to show download progress for models */
+  showDownloadProgress?: boolean
+  /** Custom cache directory for model storage */
+  cacheDir?: string
+}
+
+/**
+ * Embedding model type configuration for Node.js bindings.
+ *
+ * This struct represents different embedding model sources:
+ * - `preset`: Use a named preset (e.g., "balanced", "fast", "quality", "multilingual")
+ * - `fastembed`: Use a FastEmbed model with custom dimensions
+ * - `custom`: Use a custom ONNX model
+ */
+export interface JsEmbeddingModelType {
+  /** Type of model: "preset", "fastembed", or "custom" */
+  modelType: string
+  /** For preset: preset name; for fastembed/custom: model ID */
+  value: string
+  /** Number of dimensions (only for fastembed/custom) */
+  dimensions?: number
+}
+
+export interface JsExtractedImage {
+  data: Buffer
+  format: string
+  imageIndex: number
+  pageNumber?: number
+  width?: number
+  height?: number
+  colorspace?: string
+  bitsPerComponent?: number
+  isMask: boolean
+  description?: string
+  ocrResult?: JsExtractionResult | undefined
+}
+
+export interface JsExtractionConfig {
+  useCache?: boolean
+  enableQualityProcessing?: boolean
+  ocr?: JsOcrConfig
+  forceOcr?: boolean
+  chunking?: JsChunkingConfig
+  images?: JsImageExtractionConfig
+  pdfOptions?: JsPdfConfig
+  tokenReduction?: JsTokenReductionConfig
+  languageDetection?: JsLanguageDetectionConfig
+  postprocessor?: JsPostProcessorConfig
+  keywords?: JsKeywordConfig
+  htmlOptions?: JsHtmlOptions
+  maxConcurrentExtractions?: number
+}
+
+export interface JsExtractionResult {
+  content: string
+  mimeType: string
+  metadata: Metadata
+  tables: Array<JsTable>
+  detectedLanguages?: Array<string>
+  chunks?: Array<JsChunk>
+  images?: Array<JsExtractedImage>
+}
+
+export interface JsHtmlOptions {
+  headingStyle?: string
+  listIndentType?: string
+  listIndentWidth?: number
+  bullets?: string
+  strongEmSymbol?: string
+  escapeAsterisks?: boolean
+  escapeUnderscores?: boolean
+  escapeMisc?: boolean
+  escapeAscii?: boolean
+  codeLanguage?: string
+  autolinks?: boolean
+  defaultTitle?: boolean
+  brInTables?: boolean
+  hocrSpatialTables?: boolean
+  highlightStyle?: string
+  extractMetadata?: boolean
+  whitespaceMode?: string
+  stripNewlines?: boolean
+  wrap?: boolean
+  wrapWidth?: number
+  convertAsInline?: boolean
+  subSymbol?: string
+  supSymbol?: string
+  newlineStyle?: string
+  codeBlockStyle?: string
+  keepInlineImagesIn?: Array<string>
+  encoding?: string
+  debug?: boolean
+  stripTags?: Array<string>
+  preserveTags?: Array<string>
+  preprocessing?: JsHtmlPreprocessingOptions
+}
+
+export interface JsHtmlPreprocessingOptions {
+  enabled?: boolean
+  preset?: string
+  removeNavigation?: boolean
+  removeForms?: boolean
+}
+
+export interface JsImageExtractionConfig {
+  extractImages?: boolean
+  targetDpi?: number
+  maxImageDimension?: number
+  autoAdjustDpi?: boolean
+  minDpi?: number
+  maxDpi?: number
+}
+
+export interface JsKeywordConfig {
+  algorithm?: string
+  maxKeywords?: number
+  minScore?: number
+  ngramRange?: [number, number] | undefined
+  language?: string
+  yakeParams?: JsYakeParams
+  rakeParams?: JsRakeParams
+}
+
+export interface JsLanguageDetectionConfig {
+  enabled?: boolean
+  minConfidence?: number
+  detectMultiple?: boolean
+}
+
+export interface JsOcrConfig {
+  backend: string
+  language?: string
+  tesseractConfig?: JsTesseractConfig
+}
+
+export interface JsPdfConfig {
+  extractImages?: boolean
+  passwords?: Array<string>
+  extractMetadata?: boolean
+}
+
+export interface JsPostProcessorConfig {
+  enabled?: boolean
+  enabledProcessors?: Array<string>
+  disabledProcessors?: Array<string>
+}
+
+export interface JsRakeParams {
+  minWordLength?: number
+  maxWordsPerPhrase?: number
+}
+
+export interface JsTable {
+  cells: Array<Array<string>>
+  markdown: string
+  pageNumber: number
+}
+
+export interface JsTesseractConfig {
+  psm?: number
+  enableTableDetection?: boolean
+  tesseditCharWhitelist?: string
+}
+
+export interface JsTokenReductionConfig {
+  mode?: string
+  preserveImportantWords?: boolean
+}
+
+export interface JsYakeParams {
+  windowSize?: number
+}
+
+/**
+ * List all registered document extractors.
+ *
+ * Returns an array of names of all currently registered document extractors,
+ * including built-in extractors for PDF, Office documents, images, etc.
+ *
+ * # Returns
+ *
+ * Array of document extractor names.
+ *
+ * # Example
+ *
+ * ```typescript
+ * import { listDocumentExtractors } from 'kreuzberg';
+ *
+ * const extractors = listDocumentExtractors();
+ * console.log(extractors); // ['PDFExtractor', 'ImageExtractor', ...]
+ * ```
+ */
+export declare function listDocumentExtractors(): Array<string>
+
+/**
+ * List all available embedding preset names.
+ *
+ * Returns an array of preset names that can be used with `getEmbeddingPreset`.
+ *
+ * # Returns
+ *
+ * Array of 4 preset names: ["fast", "balanced", "quality", "multilingual"]
+ *
+ * # Example
+ *
+ * ```typescript
+ * import { listEmbeddingPresets } from 'kreuzberg';
+ *
+ * const presets = listEmbeddingPresets();
+ * console.log(presets); // ['fast', 'balanced', 'quality', 'multilingual']
+ * ```
+ */
+export declare function listEmbeddingPresets(): Array<string>
+
+/**
+ * List all registered OCR backends.
+ *
+ * Returns an array of names of all currently registered OCR backends,
+ * including built-in backends like "tesseract".
+ *
+ * # Returns
+ *
+ * Array of OCR backend names.
+ *
+ * # Example
+ *
+ * ```typescript
+ * import { listOcrBackends } from 'kreuzberg';
+ *
+ * const backends = listOcrBackends();
+ * console.log(backends); // ['tesseract', 'my-custom-backend', ...]
+ * ```
+ */
+export declare function listOcrBackends(): Array<string>
+
+/** List all registered post-processors */
+export declare function listPostProcessors(): Array<string>
+
+/** List all registered validators */
+export declare function listValidators(): Array<string>
+
+/**
+ * Load extraction configuration from a file.
+ *
+ * Automatically detects the file format based on extension:
+ * - `.toml` - TOML format
+ * - `.yaml` - YAML format
+ * - `.json` - JSON format
+ *
+ * # Parameters
+ *
+ * * `file_path` - Path to the configuration file (absolute or relative)
+ *
+ * # Returns
+ *
+ * `JsExtractionConfig` object with loaded configuration.
+ *
+ * # Errors
+ *
+ * Throws an error if:
+ * - File does not exist or is not accessible
+ * - File content is not valid TOML/YAML/JSON
+ * - Configuration structure is invalid
+ *
+ * # Example
+ *
+ * ```typescript
+ * import { loadExtractionConfigFromFile } from 'kreuzberg';
+ *
+ * // Load from TOML file
+ * const config = loadExtractionConfigFromFile('kreuzberg.toml');
+ *
+ * // Load from YAML file
+ * const config2 = loadExtractionConfigFromFile('./config.yaml');
+ *
+ * // Use with extraction
+ * const result = await extractFile('document.pdf', null, config);
+ * ```
+ */
+export declare function loadExtractionConfigFromFile(filePath: string): JsExtractionConfig
+
+/**
+ * Register a custom OCR backend
+ *
+ * Registers a JavaScript OCR backend that can process images and extract text.
+ *
+ * # Arguments
+ *
+ * * `backend` - JavaScript object with the following interface:
+ *   - `name(): string` - Unique backend name
+ *   - `supportedLanguages(): string[]` - Array of supported ISO 639-2/3 language codes
+ *   - `processImage(imageBytes: string, language: string): Promise<result>` - Process image and return extraction result
+ *
+ * # Implementation Notes
+ *
+ * Due to NAPI ThreadsafeFunction limitations, the processImage function receives:
+ * - `imageBytes` as a Base64 string (first argument)
+ * - `language` as string (second argument)
+ *
+ * And must return a Promise resolving to a JSON-serializable object with:
+ * ```typescript
+ * {
+ *   content: string,
+ *   mime_type: string,  // default: "text/plain"
+ *   metadata: object,   // default: {}
+ *   tables: array       // default: []
+ * }
+ * ```
+ *
+ * # Example
+ *
+ * ```typescript
+ * import { registerOcrBackend } from '@kreuzberg/node';
+ *
+ * registerOcrBackend({
+ *   name: () => "my-ocr",
+ *   supportedLanguages: () => ["eng", "deu", "fra"],
+ *   processImage: async (imageBytes, language) => {
+ *     const buffer = Buffer.from(imageBytes, "base64");
+ *     const text = await myOcrLibrary.process(buffer, language);
+ *     return {
+ *       content: text,
+ *       mime_type: "text/plain",
+ *       metadata: { confidence: 0.95 },
+ *       tables: []
+ *     };
+ *   }
+ * });
+ * ```
+ */
+export declare function registerOcrBackend(backend: object): void
+
+/**
+ * Register a custom postprocessor
+ *
+ * Registers a JavaScript PostProcessor that will be called after extraction.
+ *
+ * # Arguments
+ *
+ * * `processor` - JavaScript object with the following interface:
+ *   - `name(): string` - Unique processor name
+ *   - `process(...args): string` - Process function that receives JSON string as args\[0\]
+ *   - `processingStage(): "early" | "middle" | "late"` - Optional processing stage
+ *
+ * # Implementation Notes
+ *
+ * Due to NAPI ThreadsafeFunction limitations, the process function receives the extraction
+ * result as a JSON string in args\[0\] and must return a JSON string. Use the TypeScript
+ * wrapper functions for a cleaner API.
+ *
+ * # Example
+ *
+ * ```typescript
+ * import { registerPostProcessor } from '@kreuzberg/node';
+ *
+ * registerPostProcessor({
+ *   name: () => "word-counter",
+ *   processingStage: () => "middle",
+ *   process: (...args) => {
+ *     const result = JSON.parse(args[0]);
+ *     const wordCount = result.content.split(/\s+/).length;
+ *     result.metadata.word_count = wordCount;
+ *     return JSON.stringify(result);
+ *   }
+ * });
+ * ```
+ */
+export declare function registerPostProcessor(processor: object): void
+
+/**
+ * Register a custom validator
+ *
+ * Registers a JavaScript Validator that will be called after extraction.
+ *
+ * # Arguments
+ *
+ * * `validator` - JavaScript object with the following interface:
+ *   - `name(): string` - Unique validator name
+ *   - `validate(...args): Promise<string>` - Validate function that receives JSON string as args\[0\]
+ *   - `priority(): number` - Optional priority (defaults to 50, higher runs first)
+ *
+ * # Implementation Notes
+ *
+ * Due to NAPI ThreadsafeFunction limitations, the validate function receives the extraction
+ * result as a JSON string in args\[0\]. On success, return an empty string. On validation
+ * failure, throw an error (the Promise should reject). Use the TypeScript wrapper functions
+ * for a cleaner API.
+ *
+ * # Example
+ *
+ * ```typescript
+ * import { registerValidator } from '@kreuzberg/node';
+ *
+ * registerValidator({
+ *   name: () => "min-length",
+ *   priority: () => 100,
+ *   validate: async (...args) => {
+ *     const result = JSON.parse(args[0]);
+ *     if (result.content.length < 100) {
+ *       throw new Error("ValidationError: Content too short");
+ *     }
+ *     return ""; // Success - return empty string
+ *   }
+ * });
+ * ```
+ */
+export declare function registerValidator(validator: object): void
+
+/**
+ * Unregister a document extractor by name.
+ *
+ * Removes the specified document extractor from the registry. If the extractor
+ * doesn't exist, this operation is a no-op (does not throw an error).
+ *
+ * # Parameters
+ *
+ * * `name` - Name of the document extractor to unregister
+ *
+ * # Example
+ *
+ * ```typescript
+ * import { unregisterDocumentExtractor } from 'kreuzberg';
+ *
+ * // Unregister a custom extractor
+ * unregisterDocumentExtractor('MyCustomExtractor');
+ * ```
+ */
+export declare function unregisterDocumentExtractor(name: string): void
+
+/**
+ * Unregister an OCR backend by name.
+ *
+ * Removes the specified OCR backend from the registry. If the backend doesn't exist,
+ * this operation is a no-op (does not throw an error).
+ *
+ * # Parameters
+ *
+ * * `name` - Name of the OCR backend to unregister
+ *
+ * # Example
+ *
+ * ```typescript
+ * import { unregisterOcrBackend } from 'kreuzberg';
+ *
+ * // Unregister a custom backend
+ * unregisterOcrBackend('my-custom-ocr');
+ * ```
+ */
+export declare function unregisterOcrBackend(name: string): void
+
+/** Unregister a postprocessor by name */
+export declare function unregisterPostProcessor(name: string): void
+
+/** Unregister a validator by name */
+export declare function unregisterValidator(name: string): void
+
+/**
+ * Validate that a MIME type is supported by Kreuzberg.
+ *
+ * Checks if a MIME type is in the list of supported formats. Note that any
+ * `image/*` MIME type is automatically considered valid.
+ *
+ * # Parameters
+ *
+ * * `mime_type` - The MIME type to validate (string)
+ *
+ * # Returns
+ *
+ * The validated MIME type (may be normalized).
+ *
+ * # Errors
+ *
+ * Throws an error if the MIME type is not supported.
+ *
+ * # Example
+ *
+ * ```typescript
+ * import { validateMimeType } from 'kreuzberg';
+ *
+ * // Validate supported type
+ * const validated = validateMimeType('application/pdf');
+ * console.log(validated); // 'application/pdf'
+ *
+ * // Validate custom image type
+ * const validated2 = validateMimeType('image/custom-format');
+ * console.log(validated2); // 'image/custom-format' (any image/* is valid)
+ *
+ * // Validate unsupported type (throws error)
+ * try {
+ *   validateMimeType('video/mp4');
+ * } catch (err) {
+ *   console.error(err); // Error: Unsupported format: video/mp4
+ * }
+ * ```
+ */
+export declare function validateMimeType(mimeType: string): string