@kreuzberg/node 4.0.8 → 4.1.0

package/dist/index.mjs

@@ -1,20 +1,14 @@
-import { readFileSync } from "node:fs";
-import { createRequire } from "node:module";
-import {
-  CacheError,
-  ErrorCode,
-  ImageProcessingError,
-  KreuzbergError,
-  MissingDependencyError,
-  OcrError,
-  ParsingError,
-  PluginError,
-  ValidationError
-} from "./errors.js";
-import { GutenOcrBackend } from "./ocr/guten-ocr.js";
-export * from "./types.js";
-let binding = null;
-let bindingInitialized = false;
+var __require = /* @__PURE__ */ ((x) => typeof require !== "undefined" ? require : typeof Proxy !== "undefined" ? new Proxy(x, {
+  get: (a, b) => (typeof require !== "undefined" ? require : a)[b]
+}) : x)(function(x) {
+  if (typeof require !== "undefined") return require.apply(this, arguments);
+  throw Error('Dynamic require of "' + x + '" is not supported');
+});
+
+// typescript/core/binding.ts
+import { createRequire } from "module";
+var binding = null;
+var bindingInitialized = false;
 function createNativeBindingError(error) {
   const hintParts = [];
   let detail = "Unknown error while requiring native module.";
@@ -43,42 +37,13 @@ function createNativeBindingError(error) {
     ].join(" ")
   );
 }
-function assertUint8Array(value, name) {
-  if (!(value instanceof Uint8Array)) {
-    throw new TypeError(`${name} must be a Uint8Array`);
-  }
-  return value;
-}
-function assertUint8ArrayList(values, name) {
-  if (!Array.isArray(values)) {
-    throw new TypeError(`${name} must be an array of Uint8Array`);
-  }
-  const array = values;
-  return array.map((value, index) => {
-    try {
-      return assertUint8Array(value, `${name}[${index}]`);
-    } catch {
-      throw new TypeError(`${name}[${index}] must be a Uint8Array`);
-    }
-  });
-}
-function __setBindingForTests(mock) {
-  binding = mock;
-  bindingInitialized = true;
-}
-function __resetBindingForTests() {
-  binding = null;
-  bindingInitialized = false;
-}
 function loadNativeBinding() {
   let localRequire;
-  if (typeof require !== "undefined") {
-    localRequire = require;
-  } else {
-    try {
-      localRequire = createRequire(import.meta.url);
-    } catch {
-      localRequire = void 0;
+  try {
+    localRequire = createRequire(import.meta.url);
+  } catch {
+    if (typeof __require !== "undefined") {
+      localRequire = __require;
     }
   }
   if (!localRequire) {
@@ -131,175 +96,158 @@ function getBinding() {
     "Failed to load Kreuzberg bindings. Neither NAPI (Node.js) nor WASM (browsers/Deno) bindings are available. Make sure you have installed the @kreuzberg/node package for Node.js/Bun."
   );
 }
-function parseMetadata(metadataStr) {
-  try {
-    const parsed = JSON.parse(metadataStr);
-    if (typeof parsed === "object" && parsed !== null) {
-      return parsed;
-    }
-    return {};
-  } catch {
-    return {};
-  }
+function __setBindingForTests(mock) {
+  binding = mock;
+  bindingInitialized = true;
 }
-function ensureUint8Array(value) {
-  if (value instanceof Uint8Array) {
-    return value;
-  }
-  if (typeof Buffer !== "undefined" && value instanceof Buffer) {
-    return new Uint8Array(value);
-  }
-  if (Array.isArray(value)) {
-    return new Uint8Array(value);
-  }
-  return new Uint8Array();
+function __resetBindingForTests() {
+  binding = null;
+  bindingInitialized = false;
 }
-function convertChunk(rawChunk) {
-  if (!rawChunk || typeof rawChunk !== "object") {
-    return {
-      content: "",
-      metadata: {
-        byteStart: 0,
-        byteEnd: 0,
-        tokenCount: null,
-        chunkIndex: 0,
-        totalChunks: 0
-      },
-      embedding: null
-    };
-  }
-  const chunk = rawChunk;
-  const metadata = chunk["metadata"] ?? {};
-  return {
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
-    content: chunk["content"] ?? "",
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
-    embedding: chunk["embedding"] ?? null,
-    metadata: {
-      // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
-      byteStart: metadata["byte_start"] ?? metadata["charStart"] ?? 0,
-      // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
-      byteEnd: metadata["byte_end"] ?? metadata["charEnd"] ?? 0,
-      // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
-      tokenCount: metadata["token_count"] ?? metadata["tokenCount"] ?? null,
-      // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
-      chunkIndex: metadata["chunk_index"] ?? metadata["chunkIndex"] ?? 0,
-      // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
-      totalChunks: metadata["total_chunks"] ?? metadata["totalChunks"] ?? 0,
-      // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
-      firstPage: metadata["first_page"] ?? metadata["firstPage"] ?? null,
-      // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
-      lastPage: metadata["last_page"] ?? metadata["lastPage"] ?? null
-    }
-  };
+
+// typescript/errors/diagnostics.ts
+function getLastErrorCode() {
+  const binding2 = getBinding();
+  return binding2.getLastErrorCode();
 }
-function convertImage(rawImage) {
-  if (!rawImage || typeof rawImage !== "object") {
-    return {
-      data: new Uint8Array(),
-      format: "unknown",
-      imageIndex: 0,
-      pageNumber: null,
-      width: null,
-      height: null,
-      colorspace: null,
-      bitsPerComponent: null,
-      isMask: false,
-      description: null,
-      ocrResult: null
-    };
-  }
-  const image = rawImage;
-  return {
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
-    data: ensureUint8Array(image["data"]),
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
-    format: image["format"] ?? "unknown",
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
-    imageIndex: image["imageIndex"] ?? 0,
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
-    pageNumber: image["pageNumber"] ?? null,
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
-    width: image["width"] ?? null,
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
-    height: image["height"] ?? null,
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
-    colorspace: image["colorspace"] ?? null,
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
-    bitsPerComponent: image["bitsPerComponent"] ?? null,
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
-    isMask: image["isMask"] ?? false,
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
-    description: image["description"] ?? null,
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
-    ocrResult: image["ocrResult"] ? convertResult(image["ocrResult"]) : null
-  };
+function getLastPanicContext() {
+  const binding2 = getBinding();
+  const result = binding2.getLastPanicContext();
+  return result;
 }
-function convertPageContent(rawPage) {
-  if (!rawPage || typeof rawPage !== "object") {
+function getErrorCodeName(code) {
+  const binding2 = getBinding();
+  return binding2.getErrorCodeName(code);
+}
+function getErrorCodeDescription(code) {
+  const binding2 = getBinding();
+  return binding2.getErrorCodeDescription(code);
+}
+function classifyError(errorMessage) {
+  const binding2 = getBinding();
+  const result = binding2.classifyError(errorMessage);
+  return result;
+}
+
+// typescript/errors.ts
+var ErrorCode = /* @__PURE__ */ ((ErrorCode2) => {
+  ErrorCode2[ErrorCode2["Success"] = 0] = "Success";
+  ErrorCode2[ErrorCode2["GenericError"] = 1] = "GenericError";
+  ErrorCode2[ErrorCode2["Panic"] = 2] = "Panic";
+  ErrorCode2[ErrorCode2["InvalidArgument"] = 3] = "InvalidArgument";
+  ErrorCode2[ErrorCode2["IoError"] = 4] = "IoError";
+  ErrorCode2[ErrorCode2["ParsingError"] = 5] = "ParsingError";
+  ErrorCode2[ErrorCode2["OcrError"] = 6] = "OcrError";
+  ErrorCode2[ErrorCode2["MissingDependency"] = 7] = "MissingDependency";
+  return ErrorCode2;
+})(ErrorCode || {});
+var KreuzbergError = class _KreuzbergError extends Error {
+  /**
+   * Panic context if error was caused by a panic in native code.
+   * Will be null for non-panic errors.
+   */
+  panicContext;
+  constructor(message, panicContext) {
+    super(message);
+    this.name = "KreuzbergError";
+    this.panicContext = panicContext ?? null;
+    Object.setPrototypeOf(this, _KreuzbergError.prototype);
+  }
+  toJSON() {
     return {
-      pageNumber: 0,
-      content: "",
-      tables: [],
-      images: []
+      name: this.name,
+      message: this.message,
+      panicContext: this.panicContext,
+      stack: this.stack
     };
   }
-  const page = rawPage;
-  return {
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
-    pageNumber: page["pageNumber"] ?? 0,
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
-    content: page["content"] ?? "",
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
-    tables: Array.isArray(page["tables"]) ? page["tables"] : [],
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
-    images: Array.isArray(page["images"]) ? page["images"].map((image) => convertImage(image)) : []
-  };
-}
-function convertResult(rawResult) {
-  if (!rawResult || typeof rawResult !== "object") {
+};
+var ValidationError = class _ValidationError extends KreuzbergError {
+  constructor(message, panicContext) {
+    super(message, panicContext);
+    this.name = "ValidationError";
+    Object.setPrototypeOf(this, _ValidationError.prototype);
+  }
+};
+var ParsingError = class _ParsingError extends KreuzbergError {
+  constructor(message, panicContext) {
+    super(message, panicContext);
+    this.name = "ParsingError";
+    Object.setPrototypeOf(this, _ParsingError.prototype);
+  }
+};
+var OcrError = class _OcrError extends KreuzbergError {
+  constructor(message, panicContext) {
+    super(message, panicContext);
+    this.name = "OcrError";
+    Object.setPrototypeOf(this, _OcrError.prototype);
+  }
+};
+var CacheError = class _CacheError extends KreuzbergError {
+  constructor(message, panicContext) {
+    super(message, panicContext);
+    this.name = "CacheError";
+    Object.setPrototypeOf(this, _CacheError.prototype);
+  }
+};
+var ImageProcessingError = class _ImageProcessingError extends KreuzbergError {
+  constructor(message, panicContext) {
+    super(message, panicContext);
+    this.name = "ImageProcessingError";
+    Object.setPrototypeOf(this, _ImageProcessingError.prototype);
+  }
+};
+var PluginError = class _PluginError extends KreuzbergError {
+  /**
+   * Name of the plugin that threw the error.
+   */
+  pluginName;
+  constructor(message, pluginName, panicContext) {
+    super(`Plugin error in '${pluginName}': ${message}`, panicContext);
+    this.name = "PluginError";
+    this.pluginName = pluginName;
+    Object.setPrototypeOf(this, _PluginError.prototype);
+  }
+  toJSON() {
     return {
-      content: "",
-      mimeType: "application/octet-stream",
-      metadata: {},
-      tables: [],
-      detectedLanguages: null,
-      chunks: null,
-      images: null,
-      pages: null
+      name: this.name,
+      message: this.message,
+      pluginName: this.pluginName,
+      panicContext: this.panicContext,
+      stack: this.stack
     };
   }
-  const result = rawResult;
-  const metadata = result["metadata"];
-  const metadataValue = typeof metadata === "string" ? parseMetadata(metadata) : metadata ?? {};
-  const returnObj = {
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
-    content: result["content"] ?? "",
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
-    mimeType: result["mimeType"] ?? "application/octet-stream",
-    metadata: metadataValue,
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
-    tables: Array.isArray(result["tables"]) ? result["tables"] : [],
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
-    detectedLanguages: Array.isArray(result["detectedLanguages"]) ? result["detectedLanguages"] : null,
-    chunks: null,
-    images: null,
-    pages: null
-  };
-  const chunksData = result["chunks"];
-  if (Array.isArray(chunksData)) {
-    returnObj.chunks = chunksData.map((chunk) => convertChunk(chunk));
+};
+var MissingDependencyError = class _MissingDependencyError extends KreuzbergError {
+  constructor(message, panicContext) {
+    super(message, panicContext);
+    this.name = "MissingDependencyError";
+    Object.setPrototypeOf(this, _MissingDependencyError.prototype);
   }
-  const imagesData = result["images"];
-  if (Array.isArray(imagesData)) {
-    returnObj.images = imagesData.map((image) => convertImage(image));
+};
+
+// typescript/core/assertions.ts
+function assertUint8Array(value, name) {
+  if (!(value instanceof Uint8Array)) {
+    throw new TypeError(`${name} must be a Uint8Array`);
   }
-  const pagesData = result["pages"];
-  if (Array.isArray(pagesData)) {
-    returnObj.pages = pagesData.map((page) => convertPageContent(page));
+  return value;
+}
+function assertUint8ArrayList(values, name) {
+  if (!Array.isArray(values)) {
+    throw new TypeError(`${name} must be an array of Uint8Array`);
   }
-  return returnObj;
+  const array = values;
+  return array.map((value, index) => {
+    try {
+      return assertUint8Array(value, `${name}[${index}]`);
+    } catch {
+      throw new TypeError(`${name}[${index}] must be a Uint8Array`);
+    }
+  });
 }
+
+// typescript/core/config-normalizer.ts
 function setIfDefined(target, key, value) {
   if (value !== void 0) {
     target[key] = value;
@@ -457,47 +405,251 @@ function normalizeKeywordConfig(config) {
   setIfDefined(normalized, "rakeParams", config.rakeParams);
   return normalized;
 }
-function normalizePageConfig(pages) {
-  if (!pages) {
-    return void 0;
+function normalizePageConfig(pages) {
+  if (!pages) {
+    return void 0;
+  }
+  const normalized = {};
+  setIfDefined(normalized, "extractPages", pages.extractPages);
+  setIfDefined(normalized, "insertPageMarkers", pages.insertPageMarkers);
+  setIfDefined(normalized, "markerFormat", pages.markerFormat);
+  return normalized;
+}
+function normalizeExtractionConfig(config) {
+  if (!config) {
+    return null;
+  }
+  const normalized = {};
+  setIfDefined(normalized, "useCache", config.useCache);
+  setIfDefined(normalized, "enableQualityProcessing", config.enableQualityProcessing);
+  setIfDefined(normalized, "forceOcr", config.forceOcr);
+  setIfDefined(normalized, "maxConcurrentExtractions", config.maxConcurrentExtractions);
+  const ocr = normalizeOcrConfig(config.ocr);
+  setIfDefined(normalized, "ocr", ocr);
+  const chunking = normalizeChunkingConfig(config.chunking);
+  setIfDefined(normalized, "chunking", chunking);
+  const images = normalizeImageExtractionConfig(config.images);
+  setIfDefined(normalized, "images", images);
+  const pdf = normalizePdfConfig(config.pdfOptions);
+  setIfDefined(normalized, "pdfOptions", pdf);
+  const tokenReduction = normalizeTokenReductionConfig(config.tokenReduction);
+  setIfDefined(normalized, "tokenReduction", tokenReduction);
+  const languageDetection = normalizeLanguageDetectionConfig(config.languageDetection);
+  setIfDefined(normalized, "languageDetection", languageDetection);
+  const postprocessor = normalizePostProcessorConfig(config.postprocessor);
+  setIfDefined(normalized, "postprocessor", postprocessor);
+  const keywords = normalizeKeywordConfig(config.keywords);
+  setIfDefined(normalized, "keywords", keywords);
+  const pages = normalizePageConfig(config.pages);
+  setIfDefined(normalized, "pages", pages);
+  const htmlOptions = normalizeHtmlOptions(config.htmlOptions);
+  setIfDefined(normalized, "htmlOptions", htmlOptions);
+  return normalized;
+}
+
+// typescript/core/type-converters.ts
+function parseMetadata(metadataStr) {
+  try {
+    const parsed = JSON.parse(metadataStr);
+    if (typeof parsed === "object" && parsed !== null) {
+      return parsed;
+    }
+    return {};
+  } catch {
+    return {};
+  }
+}
+function ensureUint8Array(value) {
+  if (value instanceof Uint8Array) {
+    return value;
+  }
+  if (typeof Buffer !== "undefined" && value instanceof Buffer) {
+    return new Uint8Array(value);
+  }
+  if (Array.isArray(value)) {
+    return new Uint8Array(value);
+  }
+  return new Uint8Array();
+}
+function convertChunk(rawChunk) {
+  if (!rawChunk || typeof rawChunk !== "object") {
+    return {
+      content: "",
+      metadata: {
+        byteStart: 0,
+        byteEnd: 0,
+        tokenCount: null,
+        chunkIndex: 0,
+        totalChunks: 0
+      },
+      embedding: null
+    };
+  }
+  const chunk = rawChunk;
+  const metadata = chunk["metadata"] ?? {};
+  return {
+    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
+    content: chunk["content"] ?? "",
+    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
+    embedding: chunk["embedding"] ?? null,
+    metadata: {
+      // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
+      byteStart: metadata["byte_start"] ?? metadata["charStart"] ?? 0,
+      // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
+      byteEnd: metadata["byte_end"] ?? metadata["charEnd"] ?? 0,
+      // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
+      tokenCount: metadata["token_count"] ?? metadata["tokenCount"] ?? null,
+      // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
+      chunkIndex: metadata["chunk_index"] ?? metadata["chunkIndex"] ?? 0,
+      // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
+      totalChunks: metadata["total_chunks"] ?? metadata["totalChunks"] ?? 0,
+      // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
+      firstPage: metadata["first_page"] ?? metadata["firstPage"] ?? null,
+      // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
+      lastPage: metadata["last_page"] ?? metadata["lastPage"] ?? null
+    }
+  };
+}
+function convertImage(rawImage) {
+  if (!rawImage || typeof rawImage !== "object") {
+    return {
+      data: new Uint8Array(),
+      format: "unknown",
+      imageIndex: 0,
+      pageNumber: null,
+      width: null,
+      height: null,
+      colorspace: null,
+      bitsPerComponent: null,
+      isMask: false,
+      description: null,
+      ocrResult: null
+    };
+  }
+  const image = rawImage;
+  return {
+    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
+    data: ensureUint8Array(image["data"]),
+    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
+    format: image["format"] ?? "unknown",
+    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
+    imageIndex: image["imageIndex"] ?? 0,
+    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
+    pageNumber: image["pageNumber"] ?? null,
+    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
+    width: image["width"] ?? null,
+    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
+    height: image["height"] ?? null,
+    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
+    colorspace: image["colorspace"] ?? null,
+    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
+    bitsPerComponent: image["bitsPerComponent"] ?? null,
+    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
+    isMask: image["isMask"] ?? false,
+    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
+    description: image["description"] ?? null,
+    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
+    ocrResult: image["ocrResult"] ? convertResult(image["ocrResult"]) : null
+  };
+}
+function convertPageContent(rawPage) {
+  if (!rawPage || typeof rawPage !== "object") {
+    return {
+      pageNumber: 0,
+      content: "",
+      tables: [],
+      images: []
+    };
+  }
+  const page = rawPage;
+  return {
+    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
+    pageNumber: page["pageNumber"] ?? 0,
+    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
+    content: page["content"] ?? "",
+    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
+    tables: Array.isArray(page["tables"]) ? page["tables"] : [],
+    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
+    images: Array.isArray(page["images"]) ? page["images"].map((image) => convertImage(image)) : []
+  };
+}
+function convertResult(rawResult) {
+  if (!rawResult || typeof rawResult !== "object") {
+    return {
+      content: "",
+      mimeType: "application/octet-stream",
+      metadata: {},
+      tables: [],
+      detectedLanguages: null,
+      chunks: null,
+      images: null,
+      pages: null
+    };
+  }
+  const result = rawResult;
+  const metadata = result["metadata"];
+  const metadataValue = typeof metadata === "string" ? parseMetadata(metadata) : metadata ?? {};
+  const returnObj = {
+    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
+    content: result["content"] ?? "",
+    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
+    mimeType: result["mimeType"] ?? "application/octet-stream",
+    metadata: metadataValue,
+    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
+    tables: Array.isArray(result["tables"]) ? result["tables"] : [],
+    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
+    detectedLanguages: Array.isArray(result["detectedLanguages"]) ? result["detectedLanguages"] : null,
+    chunks: null,
+    images: null,
+    pages: null
+  };
+  const chunksData = result["chunks"];
+  if (Array.isArray(chunksData)) {
+    returnObj.chunks = chunksData.map((chunk) => convertChunk(chunk));
+  }
+  const imagesData = result["images"];
+  if (Array.isArray(imagesData)) {
+    returnObj.images = imagesData.map((image) => convertImage(image));
+  }
+  const pagesData = result["pages"];
+  if (Array.isArray(pagesData)) {
+    returnObj.pages = pagesData.map((page) => convertPageContent(page));
+  }
+  return returnObj;
+}
+
+// typescript/extraction/batch.ts
+function batchExtractFilesSync(paths, config = null) {
+  const normalizedConfig = normalizeExtractionConfig(config);
+  const rawResults = getBinding().batchExtractFilesSync(paths, normalizedConfig);
+  return rawResults.map(convertResult);
+}
+async function batchExtractFiles(paths, config = null) {
+  const normalizedConfig = normalizeExtractionConfig(config);
+  const rawResults = await getBinding().batchExtractFiles(paths, normalizedConfig);
+  return rawResults.map(convertResult);
+}
+function batchExtractBytesSync(dataList, mimeTypes, config = null) {
+  const buffers = assertUint8ArrayList(dataList, "dataList").map((data) => Buffer.from(data));
+  if (buffers.length !== mimeTypes.length) {
+    throw new TypeError("dataList and mimeTypes must have the same length");
   }
-  const normalized = {};
-  setIfDefined(normalized, "extractPages", pages.extractPages);
-  setIfDefined(normalized, "insertPageMarkers", pages.insertPageMarkers);
-  setIfDefined(normalized, "markerFormat", pages.markerFormat);
-  return normalized;
+  const normalizedConfig = normalizeExtractionConfig(config);
+  const rawResults = getBinding().batchExtractBytesSync(buffers, mimeTypes, normalizedConfig);
+  return rawResults.map(convertResult);
 }
-function normalizeExtractionConfig(config) {
-  if (!config) {
-    return null;
+async function batchExtractBytes(dataList, mimeTypes, config = null) {
+  const buffers = assertUint8ArrayList(dataList, "dataList").map((data) => Buffer.from(data));
+  if (buffers.length !== mimeTypes.length) {
+    throw new TypeError("dataList and mimeTypes must have the same length");
   }
-  const normalized = {};
-  setIfDefined(normalized, "useCache", config.useCache);
-  setIfDefined(normalized, "enableQualityProcessing", config.enableQualityProcessing);
-  setIfDefined(normalized, "forceOcr", config.forceOcr);
-  setIfDefined(normalized, "maxConcurrentExtractions", config.maxConcurrentExtractions);
-  const ocr = normalizeOcrConfig(config.ocr);
-  setIfDefined(normalized, "ocr", ocr);
-  const chunking = normalizeChunkingConfig(config.chunking);
-  setIfDefined(normalized, "chunking", chunking);
-  const images = normalizeImageExtractionConfig(config.images);
-  setIfDefined(normalized, "images", images);
-  const pdf = normalizePdfConfig(config.pdfOptions);
-  setIfDefined(normalized, "pdfOptions", pdf);
-  const tokenReduction = normalizeTokenReductionConfig(config.tokenReduction);
-  setIfDefined(normalized, "tokenReduction", tokenReduction);
-  const languageDetection = normalizeLanguageDetectionConfig(config.languageDetection);
-  setIfDefined(normalized, "languageDetection", languageDetection);
-  const postprocessor = normalizePostProcessorConfig(config.postprocessor);
-  setIfDefined(normalized, "postprocessor", postprocessor);
-  const keywords = normalizeKeywordConfig(config.keywords);
-  setIfDefined(normalized, "keywords", keywords);
-  const pages = normalizePageConfig(config.pages);
-  setIfDefined(normalized, "pages", pages);
-  const htmlOptions = normalizeHtmlOptions(config.htmlOptions);
-  setIfDefined(normalized, "htmlOptions", htmlOptions);
-  return normalized;
+  const normalizedConfig = normalizeExtractionConfig(config);
+  const rawResults = await getBinding().batchExtractBytes(buffers, mimeTypes, normalizedConfig);
+  return rawResults.map(convertResult);
 }
+
+// typescript/extraction/single.ts
+import { readFileSync } from "fs";
 function extractFileSync(filePath, mimeTypeOrConfig, maybeConfig) {
   let mimeType = null;
   let config = null;
@@ -559,34 +711,57 @@ async function extractBytes(dataOrPath, mimeType, config = null) {
   const rawResult = await getBinding().extractBytes(Buffer.from(validated), mimeType, normalizedConfig);
   return convertResult(rawResult);
 }
-function batchExtractFilesSync(paths, config = null) {
-  const normalizedConfig = normalizeExtractionConfig(config);
-  const rawResults = getBinding().batchExtractFilesSync(paths, normalizedConfig);
-  return rawResults.map(convertResult);
+
+// typescript/extraction/worker-pool.ts
+function createWorkerPool(size) {
+  const binding2 = getBinding();
+  const rawPool = binding2.createWorkerPool(size);
+  return rawPool;
 }
-async function batchExtractFiles(paths, config = null) {
-  const normalizedConfig = normalizeExtractionConfig(config);
-  const rawResults = await getBinding().batchExtractFiles(paths, normalizedConfig);
-  return rawResults.map(convertResult);
+function getWorkerPoolStats(pool) {
+  const binding2 = getBinding();
+  const rawStats = binding2.getWorkerPoolStats(pool);
+  return rawStats;
 }
-function batchExtractBytesSync(dataList, mimeTypes, config = null) {
-  const buffers = assertUint8ArrayList(dataList, "dataList").map((data) => Buffer.from(data));
-  if (buffers.length !== mimeTypes.length) {
-    throw new TypeError("dataList and mimeTypes must have the same length");
+async function extractFileInWorker(pool, filePath, mimeTypeOrConfig, maybeConfig) {
+  let mimeType = null;
+  let config = null;
+  if (typeof mimeTypeOrConfig === "string") {
+    mimeType = mimeTypeOrConfig;
+    config = maybeConfig ?? null;
+  } else if (mimeTypeOrConfig !== null && typeof mimeTypeOrConfig === "object") {
+    config = mimeTypeOrConfig;
+    mimeType = null;
+  } else {
+    config = maybeConfig ?? null;
+    mimeType = null;
   }
   const normalizedConfig = normalizeExtractionConfig(config);
-  const rawResults = getBinding().batchExtractBytesSync(buffers, mimeTypes, normalizedConfig);
-  return rawResults.map(convertResult);
+  const binding2 = getBinding();
+  const rawResult = await binding2.extractFileInWorker(
+    pool,
+    filePath,
+    mimeType,
+    normalizedConfig
+  );
+  return convertResult(rawResult);
 }
-async function batchExtractBytes(dataList, mimeTypes, config = null) {
-  const buffers = assertUint8ArrayList(dataList, "dataList").map((data) => Buffer.from(data));
-  if (buffers.length !== mimeTypes.length) {
-    throw new TypeError("dataList and mimeTypes must have the same length");
-  }
+async function batchExtractFilesInWorker(pool, paths, config = null) {
   const normalizedConfig = normalizeExtractionConfig(config);
-  const rawResults = await getBinding().batchExtractBytes(buffers, mimeTypes, normalizedConfig);
+  const binding2 = getBinding();
+  const rawResults = await binding2.batchExtractFilesInWorker(
+    pool,
+    paths,
+    normalizedConfig
+  );
   return rawResults.map(convertResult);
 }
+async function closeWorkerPool(pool) {
+  const binding2 = getBinding();
+  await binding2.closeWorkerPool(pool);
+}
+
+// typescript/plugins/post-processors.ts
 function registerPostProcessor(processor) {
   const binding2 = getBinding();
   const wrappedProcessor = {
@@ -641,6 +816,8 @@ function listPostProcessors() {
   const binding2 = getBinding();
   return binding2.listPostProcessors();
 }
+
+// typescript/plugins/validators.ts
 function registerValidator(validator) {
   const binding2 = getBinding();
   const wrappedValidator = {
@@ -679,6 +856,204 @@ function listValidators() {
   const binding2 = getBinding();
   return binding2.listValidators();
 }
+
+// typescript/ocr/guten-ocr.ts
+var GutenOcrBackend = class {
+  ocr = null;
+  ocrModule = null;
+  options;
+  /**
+   * Create a new Guten OCR backend.
+   *
+   * @param options - Optional configuration for Guten OCR
+   * @param options.models - Custom model paths (default: uses bundled models)
+   * @param options.isDebug - Enable debug mode (default: false)
+   * @param options.debugOutputDir - Directory for debug output (default: undefined)
+   * @param options.onnxOptions - Custom ONNX Runtime options (default: undefined)
+   *
+   * @example
+   * ```typescript
+   * // Default configuration
+   * const backend = new GutenOcrBackend();
+   *
+   * // With debug enabled
+   * const debugBackend = new GutenOcrBackend({
+   *   isDebug: true,
+   *   debugOutputDir: './ocr_debug'
+   * });
+   * ```
+   */
+  constructor(options) {
+    if (options !== void 0) {
+      this.options = options;
+    }
+  }
+  /**
+   * Get the backend name.
+   *
+   * @returns Backend name ("guten-ocr")
+   */
+  name() {
+    return "guten-ocr";
+  }
+  /**
+   * Get list of supported language codes.
+   *
+   * Guten OCR supports multiple languages depending on the model configuration.
+   * The default models support English and Chinese.
+   *
+   * @returns Array of ISO 639-1/2 language codes
+   */
+  supportedLanguages() {
+    return ["en", "eng", "ch_sim", "ch_tra", "chinese"];
+  }
+  /**
+   * Initialize the OCR backend.
+   *
+   * This method loads the Guten OCR module and creates an OCR instance.
+   * Call this before using processImage().
+   *
+   * @throws {Error} If @gutenye/ocr-node is not installed
+   * @throws {Error} If OCR initialization fails
+   *
+   * @example
+   * ```typescript
+   * const backend = new GutenOcrBackend();
+   * await backend.initialize();
+   * ```
+   */
+  async initialize() {
+    if (this.ocr !== null) {
+      return;
+    }
+    try {
+      this.ocrModule = await import("@gutenye/ocr-node").then((m) => m.default || m);
+    } catch (e) {
+      const error = e;
+      throw new Error(
+        `Guten OCR support requires the '@gutenye/ocr-node' package. Install with: npm install @gutenye/ocr-node. Error: ${error.message}`
+      );
+    }
+    try {
+      this.ocr = await this.ocrModule?.create(this.options) ?? null;
+    } catch (e) {
+      const error = e;
+      throw new Error(`Failed to initialize Guten OCR: ${error.message}`);
+    }
+  }
+  /**
+   * Shutdown the backend and release resources.
+   *
+   * This method cleans up all resources associated with the backend,
+   * including the GutenOCR instance and module references.
+   *
+   * @example
+   * ```typescript
+   * const backend = new GutenOcrBackend();
+   * await backend.initialize();
+   * // ... use backend ...
+   * await backend.shutdown();
+   * ```
+   */
+  async shutdown() {
+    if (this.ocr !== null) {
+      this.ocr = null;
+    }
+    if (this.ocrModule !== null) {
+      this.ocrModule = null;
+    }
+  }
+  /**
+   * Process image bytes and extract text using Guten OCR.
+   *
+   * This method:
+   * 1. Decodes the image using sharp (if pixel data is needed) or passes bytes directly
+   * 2. Runs OCR detection to find text regions
+   * 3. Runs OCR recognition on each text region
+   * 4. Returns extracted text with metadata
+   *
+   * @param imageBytes - Raw image data (PNG, JPEG, TIFF, etc.)
+   * @param language - Language code (must be in supportedLanguages())
+   * @returns Promise resolving to OCR result with content and metadata
+   *
+   * @throws {Error} If backend is not initialized
+   * @throws {Error} If OCR processing fails
+   *
+   * @example
+   * ```typescript
+   * import { readFile } from 'fs/promises';
+   *
+   * const backend = new GutenOcrBackend();
+   * await backend.initialize();
+   *
+   * const imageBytes = await readFile('scanned.png');
+   * const result = await backend.processImage(imageBytes, 'en');
+   * console.log(result.content);
+   * console.log(result.metadata.confidence);
+   * ```
+   */
+  async processImage(imageBytes, language) {
+    if (this.ocr === null) {
+      await this.initialize();
+    }
+    if (this.ocr === null) {
+      throw new Error("Guten OCR backend failed to initialize");
+    }
+    try {
+      const buffer = typeof imageBytes === "string" ? Buffer.from(imageBytes, "base64") : Buffer.from(imageBytes);
+      const debugEnv = process.env["KREUZBERG_DEBUG_GUTEN"];
+      if (debugEnv === "1") {
+        const header = Array.from(buffer.subarray(0, 8));
+        console.log("[Guten OCR] Debug input header:", header);
+        console.log(
+          "[Guten OCR] Buffer?",
+          Buffer.isBuffer(buffer),
+          "constructor",
+          imageBytes?.constructor?.name,
+          "length",
+          buffer.length,
+          "type",
+          typeof imageBytes
+        );
+      }
+      let width = 0;
+      let height = 0;
+      try {
+        const sharpModule = await import("sharp");
+        const sharp = sharpModule.default || sharpModule;
+        const image = sharp(buffer);
+        const metadata = await image.metadata();
+        const metadataRecord = metadata;
+        width = metadataRecord["width"] ?? 0;
+        height = metadataRecord["height"] ?? 0;
+      } catch (metadataError) {
+        const error = metadataError;
+        console.warn(`[Guten OCR] Unable to read image metadata via sharp: ${error.message}`);
+      }
+      const result = await this.ocr.detect(buffer);
+      const textLines = result.map((line) => line.text);
+      const content = textLines.join("\n");
+      const avgConfidence = result.length > 0 ? result.reduce((sum, line) => sum + line.mean, 0) / result.length : 0;
+      return {
+        content,
+        mime_type: "text/plain",
+        metadata: {
+          width,
+          height,
+          confidence: avgConfidence,
+          text_regions: result.length,
+          language
+        },
+        tables: []
+      };
+    } catch (e) {
+      const error = e;
+      throw new Error(`Guten OCR processing failed: ${error.message}`);
+    }
+  }
+};
+
+// typescript/plugins/ocr-backends.ts
 function isOcrProcessTuple(value) {
   return Array.isArray(value) && value.length === 2 && typeof value[1] === "string" && (typeof value[0] === "string" || Buffer.isBuffer(value[0]) || value[0] instanceof Uint8Array);
 }
@@ -748,6 +1123,8 @@ function clearOcrBackends() {
   const binding2 = getBinding();
   binding2.clearOcrBackends();
 }
+
+// typescript/registry/document-extractors.ts
 function listDocumentExtractors() {
   const binding2 = getBinding();
   return binding2.listDocumentExtractors();
@@ -760,7 +1137,9 @@ function clearDocumentExtractors() {
   const binding2 = getBinding();
   binding2.clearDocumentExtractors();
 }
-const ExtractionConfig = {
+
+// typescript/config/loader.ts
+var ExtractionConfig = {
   /**
    * Load extraction configuration from a file.
    *
@@ -822,6 +1201,18 @@ const ExtractionConfig = {
     return binding2.discoverExtractionConfig();
   }
 };
+function loadConfigFile(filePath) {
+  return ExtractionConfig.fromFile(filePath);
+}
+function loadConfigFromPath(path) {
+  try {
+    return ExtractionConfig.fromFile(path);
+  } catch {
+    return ExtractionConfig.discover();
+  }
+}
+
+// typescript/mime/utilities.ts
 function detectMimeType(bytes) {
   const binding2 = getBinding();
   return binding2.detectMimeTypeFromBytes(bytes);
@@ -838,6 +1229,8 @@ function getExtensionsForMime(mimeType) {
   const binding2 = getBinding();
   return binding2.getExtensionsForMime(mimeType);
 }
+
+// typescript/embeddings/presets.ts
 function listEmbeddingPresets() {
   const binding2 = getBinding();
   return binding2.listEmbeddingPresets();
@@ -847,76 +1240,9 @@ function getEmbeddingPreset(name) {
   const result = binding2.getEmbeddingPreset(name);
   return result;
 }
-function getLastErrorCode() {
-  const binding2 = getBinding();
-  return binding2.getLastErrorCode();
-}
-function getLastPanicContext() {
-  const binding2 = getBinding();
-  const result = binding2.getLastPanicContext();
-  return result;
-}
-function getErrorCodeName(code) {
-  const binding2 = getBinding();
-  return binding2.getErrorCodeName(code);
-}
-function getErrorCodeDescription(code) {
-  const binding2 = getBinding();
-  return binding2.getErrorCodeDescription(code);
-}
-function classifyError(errorMessage) {
-  const binding2 = getBinding();
-  const result = binding2.classifyError(errorMessage);
-  return result;
-}
-function createWorkerPool(size) {
-  const binding2 = getBinding();
-  const rawPool = binding2.createWorkerPool(size);
-  return rawPool;
-}
-function getWorkerPoolStats(pool) {
-  const binding2 = getBinding();
-  const rawStats = binding2.getWorkerPoolStats(pool);
-  return rawStats;
-}
-async function extractFileInWorker(pool, filePath, mimeTypeOrConfig, maybeConfig) {
-  let mimeType = null;
-  let config = null;
-  if (typeof mimeTypeOrConfig === "string") {
-    mimeType = mimeTypeOrConfig;
-    config = maybeConfig ?? null;
-  } else if (mimeTypeOrConfig !== null && typeof mimeTypeOrConfig === "object") {
-    config = mimeTypeOrConfig;
-    mimeType = null;
-  } else {
-    config = maybeConfig ?? null;
-    mimeType = null;
-  }
-  const normalizedConfig = normalizeExtractionConfig(config);
-  const binding2 = getBinding();
-  const rawResult = await binding2.extractFileInWorker(
-    pool,
-    filePath,
-    mimeType,
-    normalizedConfig
-  );
-  return convertResult(rawResult);
-}
-async function batchExtractFilesInWorker(pool, paths, config = null) {
-  const normalizedConfig = normalizeExtractionConfig(config);
-  const binding2 = getBinding();
-  const rawResults = await binding2.batchExtractFilesInWorker(
-    pool,
-    paths,
-    normalizedConfig
-  );
-  return rawResults.map(convertResult);
-}
-async function closeWorkerPool(pool) {
-  const binding2 = getBinding();
-  await binding2.closeWorkerPool(pool);
-}
-const __version__ = "4.0.8";
+
+// typescript/index.ts
+var __version__ = "4.1.0";
 export {
   CacheError,
   ErrorCode,
@@ -963,6 +1289,8 @@ export {
   listOcrBackends,
   listPostProcessors,
   listValidators,
+  loadConfigFile,
+  loadConfigFromPath,
   registerOcrBackend,
   registerPostProcessor,
   registerValidator,