@kreuzberg/node 4.0.0-rc.15 → 4.0.0-rc.17

package/dist/index.js

@@ -37,6 +37,7 @@ __export(index_exports, {
   batchExtractBytesSync: () => batchExtractBytesSync,
   batchExtractFiles: () => batchExtractFiles,
   batchExtractFilesSync: () => batchExtractFilesSync,
+  classifyError: () => classifyError,
   clearDocumentExtractors: () => clearDocumentExtractors,
   clearOcrBackends: () => clearOcrBackends,
   clearPostProcessors: () => clearPostProcessors,
@@ -48,6 +49,8 @@ __export(index_exports, {
   extractFile: () => extractFile,
   extractFileSync: () => extractFileSync,
   getEmbeddingPreset: () => getEmbeddingPreset,
+  getErrorCodeDescription: () => getErrorCodeDescription,
+  getErrorCodeName: () => getErrorCodeName,
   getExtensionsForMime: () => getExtensionsForMime,
   getLastErrorCode: () => getLastErrorCode,
   getLastPanicContext: () => getLastPanicContext,
@@ -67,6 +70,7 @@ __export(index_exports, {
 });
 module.exports = __toCommonJS(index_exports);
 var import_node_module = require("node:module");
+var import_node_fs = require("node:fs");
 var import_errors = require("./errors.js");
 var import_guten_ocr = require("./ocr/guten-ocr.js");
 __reExport(index_exports, require("./types.js"), module.exports);
@@ -313,15 +317,15 @@ function convertResult(rawResult) {
       metadata: {},
       tables: [],
       detectedLanguages: null,
-      chunks: null,
-      images: null,
-      pages: null
+      chunks: void 0,
+      images: void 0,
+      pages: void 0
     };
   }
   const result = rawResult;
   const metadata = result["metadata"];
   const metadataValue = typeof metadata === "string" ? parseMetadata(metadata) : metadata ?? {};
-  return {
+  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
@@ -331,19 +335,23 @@ function convertResult(rawResult) {
     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: (() => {
-      const chunksData = result["chunks"];
-      return Array.isArray(chunksData) ? chunksData.map((chunk) => convertChunk(chunk)) : null;
-    })(),
-    images: (() => {
-      const imagesData = result["images"];
-      return Array.isArray(imagesData) ? imagesData.map((image) => convertImage(image)) : null;
-    })(),
-    pages: (() => {
-      const pagesData = result["pages"];
-      return Array.isArray(pagesData) ? pagesData.map((page) => convertPageContent(page)) : null;
-    })()
+    chunks: void 0,
+    images: void 0,
+    pages: void 0
   };
+  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) {
@@ -543,23 +551,59 @@ 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)));
@@ -599,8 +643,8 @@ 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];
@@ -653,8 +697,8 @@ 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") {
@@ -703,8 +747,8 @@ 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") {
@@ -769,7 +813,99 @@ function clearDocumentExtractors() {
   const binding2 = getBinding();
   binding2.clearDocumentExtractors();
 }
+class ExtractionConfigBuilder {
+  config = {};
+  /**
+   * Create a new builder with default configuration.
+   */
+  static default() {
+    return new ExtractionConfigBuilder();
+  }
+  /**
+   * Set OCR configuration.
+   */
+  withOcr(ocr) {
+    this.config["ocr"] = ocr;
+    return this;
+  }
+  /**
+   * Set chunking configuration.
+   */
+  withChunking(chunking) {
+    this.config["chunking"] = chunking;
+    return this;
+  }
+  /**
+   * Set image extraction configuration.
+   */
+  withImageExtraction(images) {
+    this.config["imageExtraction"] = images;
+    return this;
+  }
+  /**
+   * Set PDF configuration.
+   */
+  withPdf(pdf) {
+    this.config["pdf"] = pdf;
+    return this;
+  }
+  /**
+   * Set keyword extraction configuration.
+   */
+  withKeywords(keywords) {
+    this.config["keywords"] = keywords;
+    return this;
+  }
+  /**
+   * Set language detection configuration.
+   */
+  withLanguageDetection(languageDetection) {
+    this.config["languageDetection"] = languageDetection;
+    return this;
+  }
+  /**
+   * Set whether to enable metadata extraction.
+   */
+  withMetadataExtraction(enabled) {
+    this.config["metadataExtraction"] = enabled;
+    return this;
+  }
+  /**
+   * Set whether to enable quality mode.
+   */
+  withQualityMode(enabled) {
+    this.config["qualityMode"] = enabled;
+    return this;
+  }
+  /**
+   * Build and return the final ExtractionConfig object.
+   */
+  build() {
+    return this.config;
+  }
+}
 const ExtractionConfig = {
+  /**
+   * Create a default extraction configuration using the builder pattern.
+   *
+   * Returns a builder object that allows you to configure extraction settings
+   * using method chaining.
+   *
+   * @returns ExtractionConfigBuilder for chaining configuration calls
+   *
+   * @example
+   * ```typescript
+   * import { ExtractionConfig } from '@kreuzberg/node';
+   *
+   * const config = ExtractionConfig.default()
+   *   .withChunking({ maxChars: 2048 })
+   *   .withOcr({ backend: 'tesseract', language: 'eng' })
+   *   .build();
+   * ```
+   */
+  default() {
+    return ExtractionConfigBuilder.default();
+  },
   /**
    * Load extraction configuration from a file.
    *
@@ -835,9 +971,9 @@ function detectMimeType(bytes) {
   const binding2 = getBinding();
   return binding2.detectMimeTypeFromBytes(bytes);
 }
-function detectMimeTypeFromPath(path) {
+function detectMimeTypeFromPath(filePath, checkExists) {
   const binding2 = getBinding();
-  return binding2.detectMimeTypeFromPath(path);
+  return binding2.detectMimeTypeFromPath(filePath, checkExists);
 }
 function validateMimeType(mimeType) {
   const binding2 = getBinding();
@@ -865,7 +1001,20 @@ function getLastPanicContext() {
   const result = binding2.getLastPanicContext();
   return result;
 }
-const __version__ = "4.0.0-rc.15";
+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;
+}
+const __version__ = "4.0.0-rc.17";
 // Annotate the CommonJS export names for ESM import in node:
 0 && (module.exports = {
   CacheError,
@@ -886,6 +1035,7 @@ const __version__ = "4.0.0-rc.15";
   batchExtractBytesSync,
   batchExtractFiles,
   batchExtractFilesSync,
+  classifyError,
   clearDocumentExtractors,
   clearOcrBackends,
   clearPostProcessors,
@@ -897,6 +1047,8 @@ const __version__ = "4.0.0-rc.15";
   extractFile,
   extractFileSync,
   getEmbeddingPreset,
+  getErrorCodeDescription,
+  getErrorCodeName,
   getExtensionsForMime,
   getLastErrorCode,
   getLastPanicContext,