@kreuzberg/node 4.0.0-rc.21 → 4.0.0-rc.24

package/dist/index.mjs

@@ -71,7 +71,16 @@ function __resetBindingForTests() {
   bindingInitialized = false;
 }
 function loadNativeBinding() {
-  const localRequire = typeof require !== "undefined" ? require : createRequire(import.meta.url);
+  let localRequire;
+  if (typeof require !== "undefined") {
+    localRequire = require;
+  } else {
+    try {
+      localRequire = createRequire(import.meta.url);
+    } catch {
+      localRequire = void 0;
+    }
+  }
   if (!localRequire) {
     throw new Error("Unable to resolve native binding loader (require not available).");
   }
@@ -255,9 +264,9 @@ function convertResult(rawResult) {
       metadata: {},
       tables: [],
       detectedLanguages: null,
-      chunks: void 0,
-      images: void 0,
-      pages: void 0
+      chunks: null,
+      images: null,
+      pages: null
     };
   }
   const result = rawResult;
@@ -273,9 +282,9 @@ 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: void 0,
-    images: void 0,
-    pages: void 0
+    chunks: null,
+    images: null,
+    pages: null
   };
   const chunksData = result["chunks"];
   if (Array.isArray(chunksData)) {
@@ -453,9 +462,9 @@ 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) {
@@ -751,99 +760,7 @@ 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.
    *
@@ -952,7 +869,54 @@ function classifyError(errorMessage) {
   const result = binding2.classifyError(errorMessage);
   return result;
 }
-const __version__ = "4.0.0-rc.21";
+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.24";
 export {
   CacheError,
   ErrorCode,
@@ -971,17 +935,21 @@ export {
   batchExtractBytes,
   batchExtractBytesSync,
   batchExtractFiles,
+  batchExtractFilesInWorker,
   batchExtractFilesSync,
   classifyError,
   clearDocumentExtractors,
   clearOcrBackends,
   clearPostProcessors,
   clearValidators,
+  closeWorkerPool,
+  createWorkerPool,
   detectMimeType,
   detectMimeTypeFromPath,
   extractBytes,
   extractBytesSync,
   extractFile,
+  extractFileInWorker,
   extractFileSync,
   getEmbeddingPreset,
   getErrorCodeDescription,
@@ -989,6 +957,7 @@ export {
   getExtensionsForMime,
   getLastErrorCode,
   getLastPanicContext,
+  getWorkerPoolStats,
   listDocumentExtractors,
   listEmbeddingPresets,
   listOcrBackends,