@kreuzberg/node 4.0.0-rc.8 → 4.0.0

package/dist/index.js

@@ -36,21 +36,29 @@ __export(index_exports, {
   batchExtractBytes: () => batchExtractBytes,
   batchExtractBytesSync: () => batchExtractBytesSync,
   batchExtractFiles: () => batchExtractFiles,
+  batchExtractFilesInWorker: () => batchExtractFilesInWorker,
   batchExtractFilesSync: () => batchExtractFilesSync,
+  classifyError: () => classifyError,
   clearDocumentExtractors: () => clearDocumentExtractors,
   clearOcrBackends: () => clearOcrBackends,
   clearPostProcessors: () => clearPostProcessors,
   clearValidators: () => clearValidators,
+  closeWorkerPool: () => closeWorkerPool,
+  createWorkerPool: () => createWorkerPool,
   detectMimeType: () => detectMimeType,
   detectMimeTypeFromPath: () => detectMimeTypeFromPath,
   extractBytes: () => extractBytes,
   extractBytesSync: () => extractBytesSync,
   extractFile: () => extractFile,
+  extractFileInWorker: () => extractFileInWorker,
   extractFileSync: () => extractFileSync,
   getEmbeddingPreset: () => getEmbeddingPreset,
+  getErrorCodeDescription: () => getErrorCodeDescription,
+  getErrorCodeName: () => getErrorCodeName,
   getExtensionsForMime: () => getExtensionsForMime,
   getLastErrorCode: () => getLastErrorCode,
   getLastPanicContext: () => getLastPanicContext,
+  getWorkerPoolStats: () => getWorkerPoolStats,
   listDocumentExtractors: () => listDocumentExtractors,
   listEmbeddingPresets: () => listEmbeddingPresets,
   listOcrBackends: () => listOcrBackends,
@@ -66,6 +74,7 @@ __export(index_exports, {
   validateMimeType: () => validateMimeType
 });
 module.exports = __toCommonJS(index_exports);
+var import_node_fs = require("node:fs");
 var import_node_module = require("node:module");
 var import_errors = require("./errors.js");
 var import_guten_ocr = require("./ocr/guten-ocr.js");
@@ -129,17 +138,50 @@ function __resetBindingForTests() {
   bindingInitialized = false;
 }
 function loadNativeBinding() {
-  const localRequire = typeof require !== "undefined" ? (
-    // biome-ignore lint/suspicious/noExplicitAny: Node typings are available at runtime
-    require
-  ) : (0, import_node_module.createRequire)(import_meta.url);
+  let localRequire;
+  if (typeof require !== "undefined") {
+    localRequire = require;
+  } else {
+    try {
+      localRequire = (0, import_node_module.createRequire)(import_meta.url);
+    } catch {
+      localRequire = void 0;
+    }
+  }
   if (!localRequire) {
     throw new Error("Unable to resolve native binding loader (require not available).");
   }
-  return localRequire("../index.js");
+  const loadedModule = localRequire("../index.js");
+  if (typeof loadedModule !== "object" || loadedModule === null) {
+    throw new Error(
+      "Native binding is not a valid object. Ensure the native module is properly built and compatible."
+    );
+  }
+  const module2 = loadedModule;
+  const requiredMethods = [
+    "extractFileSync",
+    "extractFile",
+    "extractBytesSync",
+    "extractBytes",
+    "batchExtractFilesSync",
+    "batchExtractFiles",
+    "batchExtractBytesSync",
+    "batchExtractBytes"
+  ];
+  for (const method of requiredMethods) {
+    if (typeof module2[method] !== "function") {
+      throw new Error(
+        `Native binding is missing required method: ${method}. Ensure the native module is properly built and compatible.`
+      );
+    }
+  }
+  return module2;
 }
 function getBinding() {
   if (bindingInitialized) {
+    if (binding === null) {
+      throw new Error("Native binding was previously failed to load.");
+    }
     return binding;
   }
   try {
@@ -149,6 +191,7 @@ function getBinding() {
       return binding;
     }
   } catch (error) {
+    bindingInitialized = true;
     throw createNativeBindingError(error);
   }
   throw new Error(
@@ -157,7 +200,11 @@ function getBinding() {
 }
 function parseMetadata(metadataStr) {
   try {
-    return JSON.parse(metadataStr);
+    const parsed = JSON.parse(metadataStr);
+    if (typeof parsed === "object" && parsed !== null) {
+      return parsed;
+    }
+    return {};
   } catch {
     return {};
   }
@@ -175,7 +222,7 @@ function ensureUint8Array(value) {
   return new Uint8Array();
 }
 function convertChunk(rawChunk) {
-  if (!rawChunk) {
+  if (!rawChunk || typeof rawChunk !== "object") {
     return {
       content: "",
       metadata: {
@@ -188,23 +235,33 @@ function convertChunk(rawChunk) {
       embedding: null
     };
   }
-  const metadata = rawChunk.metadata ?? {};
+  const chunk = rawChunk;
+  const metadata = chunk["metadata"] ?? {};
   return {
-    content: rawChunk.content ?? "",
-    embedding: rawChunk.embedding ?? null,
+    // 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: {
-      byteStart: metadata.byte_start ?? metadata.charStart ?? 0,
-      byteEnd: metadata.byte_end ?? metadata.charEnd ?? 0,
-      tokenCount: metadata.token_count ?? metadata.tokenCount ?? null,
-      chunkIndex: metadata.chunk_index ?? metadata.chunkIndex ?? 0,
-      totalChunks: metadata.total_chunks ?? metadata.totalChunks ?? 0,
-      firstPage: metadata.first_page ?? metadata.firstPage ?? null,
-      lastPage: metadata.last_page ?? metadata.lastPage ?? null
+      // 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) {
+  if (!rawImage || typeof rawImage !== "object") {
     return {
       data: new Uint8Array(),
       format: "unknown",
@@ -219,31 +276,97 @@ function convertImage(rawImage) {
       ocrResult: null
     };
   }
+  const image = rawImage;
   return {
-    data: ensureUint8Array(rawImage.data),
-    format: rawImage.format ?? "unknown",
-    imageIndex: rawImage.imageIndex ?? 0,
-    pageNumber: rawImage.pageNumber ?? null,
-    width: rawImage.width ?? null,
-    height: rawImage.height ?? null,
-    colorspace: rawImage.colorspace ?? null,
-    bitsPerComponent: rawImage.bitsPerComponent ?? null,
-    isMask: rawImage.isMask ?? false,
-    description: rawImage.description ?? null,
-    ocrResult: rawImage.ocrResult ? convertResult(rawImage.ocrResult) : null
+    // 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 convertResult(rawResult) {
+function convertPageContent(rawPage) {
+  if (!rawPage || typeof rawPage !== "object") {
+    return {
+      pageNumber: 0,
+      content: "",
+      tables: [],
+      images: []
+    };
+  }
+  const page = rawPage;
   return {
-    content: rawResult.content,
-    mimeType: rawResult.mimeType,
-    metadata: typeof rawResult.metadata === "string" ? parseMetadata(rawResult.metadata) : rawResult.metadata,
-    tables: rawResult.tables || [],
-    detectedLanguages: rawResult.detectedLanguages || null,
-    chunks: Array.isArray(rawResult.chunks) ? rawResult.chunks.map((chunk) => convertChunk(chunk)) : null,
-    images: Array.isArray(rawResult.images) ? rawResult.images.map((image) => convertImage(image)) : null
+    // 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;
+}
 function setIfDefined(target, key, value) {
   if (value !== void 0) {
     target[key] = value;
@@ -406,9 +529,9 @@ function normalizePageConfig(pages) {
     return void 0;
   }
   const normalized = {};
-  setIfDefined(normalized, "extract_pages", pages.extractPages);
-  setIfDefined(normalized, "insert_page_markers", pages.insertPageMarkers);
-  setIfDefined(normalized, "marker_format", pages.markerFormat);
+  setIfDefined(normalized, "extractPages", pages.extractPages);
+  setIfDefined(normalized, "insertPageMarkers", pages.insertPageMarkers);
+  setIfDefined(normalized, "markerFormat", pages.markerFormat);
   return normalized;
 }
 function normalizeExtractionConfig(config) {
@@ -442,23 +565,59 @@ function normalizeExtractionConfig(config) {
   setIfDefined(normalized, "htmlOptions", htmlOptions);
   return normalized;
 }
-function extractFileSync(filePath, mimeType = null, config = null) {
+function extractFileSync(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 rawResult = getBinding().extractFileSync(filePath, mimeType, normalizedConfig);
   return convertResult(rawResult);
 }
-async function extractFile(filePath, mimeType = null, config = null) {
+async function extractFile(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 rawResult = await getBinding().extractFile(filePath, mimeType, normalizedConfig);
   return convertResult(rawResult);
 }
-function extractBytesSync(data, mimeType, config = null) {
+function extractBytesSync(dataOrPath, mimeType, config = null) {
+  let data;
+  if (typeof dataOrPath === "string") {
+    data = (0, import_node_fs.readFileSync)(dataOrPath);
+  } else {
+    data = dataOrPath;
+  }
   const validated = assertUint8Array(data, "data");
   const normalizedConfig = normalizeExtractionConfig(config);
   const rawResult = getBinding().extractBytesSync(Buffer.from(validated), mimeType, normalizedConfig);
   return convertResult(rawResult);
 }
-async function extractBytes(data, mimeType, config = null) {
+async function extractBytes(dataOrPath, mimeType, config = null) {
+  let data;
+  if (typeof dataOrPath === "string") {
+    data = (0, import_node_fs.readFileSync)(dataOrPath);
+  } else {
+    data = dataOrPath;
+  }
   const validated = assertUint8Array(data, "data");
   if (process.env["KREUZBERG_DEBUG_GUTEN"] === "1") {
     console.log("[TypeScript] Debug input header:", Array.from(validated.slice(0, 8)));
@@ -498,8 +657,8 @@ async function batchExtractBytes(dataList, mimeTypes, config = null) {
 function registerPostProcessor(processor) {
   const binding2 = getBinding();
   const wrappedProcessor = {
-    name: processor.name.bind(processor),
-    processingStage: processor.processingStage?.bind(processor),
+    name: typeof processor.name === "function" ? processor.name() : processor.name,
+    processingStage: typeof processor.processingStage === "function" ? processor.processingStage() : processor.processingStage,
     async process(...args) {
       const wrappedValue = args[0];
       const jsonString = wrappedValue[0];
@@ -552,8 +711,8 @@ function listPostProcessors() {
 function registerValidator(validator) {
   const binding2 = getBinding();
   const wrappedValidator = {
-    name: validator.name.bind(validator),
-    priority: validator.priority?.bind(validator),
+    name: typeof validator.name === "function" ? validator.name() : validator.name,
+    priority: typeof validator.priority === "function" ? validator.priority() : validator.priority,
     async validate(...args) {
       const jsonString = args[0];
       if (!jsonString || jsonString === "undefined") {
@@ -602,8 +761,8 @@ function describePayload(value) {
 function registerOcrBackend(backend) {
   const binding2 = getBinding();
   const wrappedBackend = {
-    name: backend.name.bind(backend),
-    supportedLanguages: backend.supportedLanguages.bind(backend),
+    name: typeof backend.name === "function" ? backend.name() : backend.name,
+    supportedLanguages: typeof backend.supportedLanguages === "function" ? backend.supportedLanguages() : backend.supportedLanguages ?? ["en"],
     async processImage(...processArgs) {
       const [imagePayload, maybeLanguage] = processArgs;
       if (process.env["KREUZBERG_DEBUG_GUTEN"] === "1") {
@@ -732,11 +891,11 @@ const ExtractionConfig = {
 };
 function detectMimeType(bytes) {
   const binding2 = getBinding();
-  return binding2.detectMimeType(bytes);
+  return binding2.detectMimeTypeFromBytes(bytes);
 }
-function detectMimeTypeFromPath(path, checkExists) {
+function detectMimeTypeFromPath(filePath, checkExists) {
   const binding2 = getBinding();
-  return binding2.detectMimeTypeFromPath(path, checkExists);
+  return binding2.detectMimeTypeFromPath(filePath, checkExists);
 }
 function validateMimeType(mimeType) {
   const binding2 = getBinding();
@@ -752,7 +911,8 @@ function listEmbeddingPresets() {
 }
 function getEmbeddingPreset(name) {
   const binding2 = getBinding();
-  return binding2.getEmbeddingPreset(name);
+  const result = binding2.getEmbeddingPreset(name);
+  return result;
 }
 function getLastErrorCode() {
   const binding2 = getBinding();
@@ -760,9 +920,70 @@ function getLastErrorCode() {
 }
 function getLastPanicContext() {
   const binding2 = getBinding();
-  return binding2.getLastPanicContext();
+  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__ = "4.0.0-rc.8";
+const __version__ = "4.0.0";
 // Annotate the CommonJS export names for ESM import in node:
 0 && (module.exports = {
   CacheError,
@@ -782,21 +1003,29 @@ const __version__ = "4.0.0-rc.8";
   batchExtractBytes,
   batchExtractBytesSync,
   batchExtractFiles,
+  batchExtractFilesInWorker,
   batchExtractFilesSync,
+  classifyError,
   clearDocumentExtractors,
   clearOcrBackends,
   clearPostProcessors,
   clearValidators,
+  closeWorkerPool,
+  createWorkerPool,
   detectMimeType,
   detectMimeTypeFromPath,
   extractBytes,
   extractBytesSync,
   extractFile,
+  extractFileInWorker,
   extractFileSync,
   getEmbeddingPreset,
+  getErrorCodeDescription,
+  getErrorCodeName,
   getExtensionsForMime,
   getLastErrorCode,
   getLastPanicContext,
+  getWorkerPoolStats,
   listDocumentExtractors,
   listEmbeddingPresets,
   listOcrBackends,