@kreuzberg/node 4.0.0-rc.14 → 4.0.0-rc.16

package/dist/index.d.mts

@@ -1,6 +1,6 @@
 import { PanicContext } from './errors.mjs';
 export { CacheError, ErrorCode, ImageProcessingError, KreuzbergError, MissingDependencyError, OcrError, ParsingError, PluginError, ValidationError } from './errors.mjs';
-import { ExtractionConfig as ExtractionConfig$1, ExtractionResult, PostProcessorProtocol, ValidatorProtocol, OcrBackendProtocol } from './types.mjs';
+import { ExtractionConfig as ExtractionConfig$1, ExtractionResult, PostProcessorProtocol, ValidatorProtocol, OcrBackendProtocol, ErrorClassification } from './types.mjs';
 export { ArchiveMetadata, Chunk, ChunkMetadata, ChunkingConfig, EmailMetadata, ErrorMetadata, ExcelMetadata, ExtractedImage, HtmlConversionOptions, HtmlMetadata, HtmlPreprocessingOptions, ImageExtractionConfig, ImageMetadata, ImagePreprocessingMetadata, KeywordAlgorithm, KeywordConfig, LanguageDetectionConfig, Metadata, OcrConfig, OcrMetadata, PageBoundary, PageConfig, PageContent, PageInfo, PageStructure, PageUnitType, PdfConfig, PdfMetadata, PostProcessorConfig, PptxMetadata, ProcessingStage, RakeParams, Table, TesseractConfig, TextMetadata, TokenReductionConfig, XmlMetadata, YakeParams } from './types.mjs';
 export { GutenOcrBackend } from './ocr/guten-ocr.mjs';
 
@@ -903,6 +903,75 @@ declare function getLastErrorCode(): number;
  * ```
  */
 declare function getLastPanicContext(): PanicContext | null;
-declare const __version__ = "4.0.0-rc.14";
+/**
+ * Returns the human-readable name for an error code.
+ *
+ * Maps numeric error codes to their string names, providing a consistent way
+ * to get error code names across all platforms.
+ *
+ * @param code - The numeric error code (0-7)
+ * @returns The error code name as a string (e.g., "validation", "ocr", "unknown")
+ *
+ * @example
+ * ```typescript
+ * import { getErrorCodeName } from '@kreuzberg/node';
+ *
+ * const name = getErrorCodeName(0);  // returns "validation"
+ * const name = getErrorCodeName(2);  // returns "ocr"
+ * const name = getErrorCodeName(99); // returns "unknown"
+ * ```
+ */
+declare function getErrorCodeName(code: number): string;
+/**
+ * Returns the description for an error code.
+ *
+ * Retrieves user-friendly descriptions of error types from the FFI layer.
+ *
+ * @param code - The numeric error code (0-7)
+ * @returns A brief description of the error type
+ *
+ * @example
+ * ```typescript
+ * import { getErrorCodeDescription } from '@kreuzberg/node';
+ *
+ * const desc = getErrorCodeDescription(0);  // returns "Input validation error"
+ * const desc = getErrorCodeDescription(4);  // returns "File system I/O error"
+ * const desc = getErrorCodeDescription(99); // returns "Unknown error code"
+ * ```
+ */
+declare function getErrorCodeDescription(code: number): string;
+/**
+ * Classifies an error message string into an error code category.
+ *
+ * This function analyzes the error message content and returns the most likely
+ * error code (0-7) based on keyword patterns. Used to programmatically classify
+ * errors for handling purposes.
+ *
+ * The classification is based on keyword matching:
+ * - **Validation (0)**: Keywords like "invalid", "validation", "schema", "required"
+ * - **Parsing (1)**: Keywords like "parsing", "corrupted", "malformed"
+ * - **Ocr (2)**: Keywords like "ocr", "tesseract", "language", "model"
+ * - **MissingDependency (3)**: Keywords like "not found", "missing", "dependency"
+ * - **Io (4)**: Keywords like "file", "disk", "read", "write", "permission"
+ * - **Plugin (5)**: Keywords like "plugin", "register", "extension"
+ * - **UnsupportedFormat (6)**: Keywords like "unsupported", "format", "mime"
+ * - **Internal (7)**: Keywords like "internal", "bug", "panic"
+ *
+ * @param errorMessage - The error message string to classify
+ * @returns An object with the classification details
+ *
+ * @example
+ * ```typescript
+ * import { classifyError } from '@kreuzberg/node';
+ *
+ * const result = classifyError("PDF file is corrupted");
+ * // Returns: { code: 1, name: "parsing", confidence: 0.95 }
+ *
+ * const result = classifyError("Tesseract not found");
+ * // Returns: { code: 3, name: "missing_dependency", confidence: 0.9 }
+ * ```
+ */
+declare function classifyError(errorMessage: string): ErrorClassification;
+declare const __version__ = "4.0.0-rc.16";
 
-export { type EmbeddingPreset, ExtractionConfig, ExtractionResult, OcrBackendProtocol, PanicContext, PostProcessorProtocol, ValidatorProtocol, __resetBindingForTests, __setBindingForTests, __version__, batchExtractBytes, batchExtractBytesSync, batchExtractFiles, batchExtractFilesSync, clearDocumentExtractors, clearOcrBackends, clearPostProcessors, clearValidators, detectMimeType, detectMimeTypeFromPath, extractBytes, extractBytesSync, extractFile, extractFileSync, getEmbeddingPreset, getExtensionsForMime, getLastErrorCode, getLastPanicContext, listDocumentExtractors, listEmbeddingPresets, listOcrBackends, listPostProcessors, listValidators, registerOcrBackend, registerPostProcessor, registerValidator, unregisterDocumentExtractor, unregisterOcrBackend, unregisterPostProcessor, unregisterValidator, validateMimeType };
+export { type EmbeddingPreset, ErrorClassification, ExtractionConfig, ExtractionResult, OcrBackendProtocol, PanicContext, PostProcessorProtocol, ValidatorProtocol, __resetBindingForTests, __setBindingForTests, __version__, batchExtractBytes, batchExtractBytesSync, batchExtractFiles, batchExtractFilesSync, classifyError, clearDocumentExtractors, clearOcrBackends, clearPostProcessors, clearValidators, detectMimeType, detectMimeTypeFromPath, extractBytes, extractBytesSync, extractFile, extractFileSync, getEmbeddingPreset, getErrorCodeDescription, getErrorCodeName, getExtensionsForMime, getLastErrorCode, getLastPanicContext, listDocumentExtractors, listEmbeddingPresets, listOcrBackends, listPostProcessors, listValidators, registerOcrBackend, registerPostProcessor, registerValidator, unregisterDocumentExtractor, unregisterOcrBackend, unregisterPostProcessor, unregisterValidator, validateMimeType };

package/dist/index.d.ts

@@ -1,6 +1,6 @@
 import { PanicContext } from './errors.js';
 export { CacheError, ErrorCode, ImageProcessingError, KreuzbergError, MissingDependencyError, OcrError, ParsingError, PluginError, ValidationError } from './errors.js';
-import { ExtractionConfig as ExtractionConfig$1, ExtractionResult, PostProcessorProtocol, ValidatorProtocol, OcrBackendProtocol } from './types.js';
+import { ExtractionConfig as ExtractionConfig$1, ExtractionResult, PostProcessorProtocol, ValidatorProtocol, OcrBackendProtocol, ErrorClassification } from './types.js';
 export { ArchiveMetadata, Chunk, ChunkMetadata, ChunkingConfig, EmailMetadata, ErrorMetadata, ExcelMetadata, ExtractedImage, HtmlConversionOptions, HtmlMetadata, HtmlPreprocessingOptions, ImageExtractionConfig, ImageMetadata, ImagePreprocessingMetadata, KeywordAlgorithm, KeywordConfig, LanguageDetectionConfig, Metadata, OcrConfig, OcrMetadata, PageBoundary, PageConfig, PageContent, PageInfo, PageStructure, PageUnitType, PdfConfig, PdfMetadata, PostProcessorConfig, PptxMetadata, ProcessingStage, RakeParams, Table, TesseractConfig, TextMetadata, TokenReductionConfig, XmlMetadata, YakeParams } from './types.js';
 export { GutenOcrBackend } from './ocr/guten-ocr.js';
 
@@ -903,6 +903,75 @@ declare function getLastErrorCode(): number;
  * ```
  */
 declare function getLastPanicContext(): PanicContext | null;
-declare const __version__ = "4.0.0-rc.14";
+/**
+ * Returns the human-readable name for an error code.
+ *
+ * Maps numeric error codes to their string names, providing a consistent way
+ * to get error code names across all platforms.
+ *
+ * @param code - The numeric error code (0-7)
+ * @returns The error code name as a string (e.g., "validation", "ocr", "unknown")
+ *
+ * @example
+ * ```typescript
+ * import { getErrorCodeName } from '@kreuzberg/node';
+ *
+ * const name = getErrorCodeName(0);  // returns "validation"
+ * const name = getErrorCodeName(2);  // returns "ocr"
+ * const name = getErrorCodeName(99); // returns "unknown"
+ * ```
+ */
+declare function getErrorCodeName(code: number): string;
+/**
+ * Returns the description for an error code.
+ *
+ * Retrieves user-friendly descriptions of error types from the FFI layer.
+ *
+ * @param code - The numeric error code (0-7)
+ * @returns A brief description of the error type
+ *
+ * @example
+ * ```typescript
+ * import { getErrorCodeDescription } from '@kreuzberg/node';
+ *
+ * const desc = getErrorCodeDescription(0);  // returns "Input validation error"
+ * const desc = getErrorCodeDescription(4);  // returns "File system I/O error"
+ * const desc = getErrorCodeDescription(99); // returns "Unknown error code"
+ * ```
+ */
+declare function getErrorCodeDescription(code: number): string;
+/**
+ * Classifies an error message string into an error code category.
+ *
+ * This function analyzes the error message content and returns the most likely
+ * error code (0-7) based on keyword patterns. Used to programmatically classify
+ * errors for handling purposes.
+ *
+ * The classification is based on keyword matching:
+ * - **Validation (0)**: Keywords like "invalid", "validation", "schema", "required"
+ * - **Parsing (1)**: Keywords like "parsing", "corrupted", "malformed"
+ * - **Ocr (2)**: Keywords like "ocr", "tesseract", "language", "model"
+ * - **MissingDependency (3)**: Keywords like "not found", "missing", "dependency"
+ * - **Io (4)**: Keywords like "file", "disk", "read", "write", "permission"
+ * - **Plugin (5)**: Keywords like "plugin", "register", "extension"
+ * - **UnsupportedFormat (6)**: Keywords like "unsupported", "format", "mime"
+ * - **Internal (7)**: Keywords like "internal", "bug", "panic"
+ *
+ * @param errorMessage - The error message string to classify
+ * @returns An object with the classification details
+ *
+ * @example
+ * ```typescript
+ * import { classifyError } from '@kreuzberg/node';
+ *
+ * const result = classifyError("PDF file is corrupted");
+ * // Returns: { code: 1, name: "parsing", confidence: 0.95 }
+ *
+ * const result = classifyError("Tesseract not found");
+ * // Returns: { code: 3, name: "missing_dependency", confidence: 0.9 }
+ * ```
+ */
+declare function classifyError(errorMessage: string): ErrorClassification;
+declare const __version__ = "4.0.0-rc.16";
 
-export { type EmbeddingPreset, ExtractionConfig, ExtractionResult, OcrBackendProtocol, PanicContext, PostProcessorProtocol, ValidatorProtocol, __resetBindingForTests, __setBindingForTests, __version__, batchExtractBytes, batchExtractBytesSync, batchExtractFiles, batchExtractFilesSync, clearDocumentExtractors, clearOcrBackends, clearPostProcessors, clearValidators, detectMimeType, detectMimeTypeFromPath, extractBytes, extractBytesSync, extractFile, extractFileSync, getEmbeddingPreset, getExtensionsForMime, getLastErrorCode, getLastPanicContext, listDocumentExtractors, listEmbeddingPresets, listOcrBackends, listPostProcessors, listValidators, registerOcrBackend, registerPostProcessor, registerValidator, unregisterDocumentExtractor, unregisterOcrBackend, unregisterPostProcessor, unregisterValidator, validateMimeType };
+export { type EmbeddingPreset, ErrorClassification, ExtractionConfig, ExtractionResult, OcrBackendProtocol, PanicContext, PostProcessorProtocol, ValidatorProtocol, __resetBindingForTests, __setBindingForTests, __version__, batchExtractBytes, batchExtractBytesSync, batchExtractFiles, batchExtractFilesSync, classifyError, clearDocumentExtractors, clearOcrBackends, clearPostProcessors, clearValidators, detectMimeType, detectMimeTypeFromPath, extractBytes, extractBytesSync, extractFile, extractFileSync, getEmbeddingPreset, getErrorCodeDescription, getErrorCodeName, getExtensionsForMime, getLastErrorCode, getLastPanicContext, listDocumentExtractors, listEmbeddingPresets, listOcrBackends, listPostProcessors, listValidators, registerOcrBackend, registerPostProcessor, registerValidator, unregisterDocumentExtractor, unregisterOcrBackend, unregisterPostProcessor, unregisterValidator, validateMimeType };

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,
@@ -865,7 +868,20 @@ function getLastPanicContext() {
   const result = binding2.getLastPanicContext();
   return result;
 }
-const __version__ = "4.0.0-rc.14";
+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.16";
 // Annotate the CommonJS export names for ESM import in node:
 0 && (module.exports = {
   CacheError,
@@ -886,6 +902,7 @@ const __version__ = "4.0.0-rc.14";
   batchExtractBytesSync,
   batchExtractFiles,
   batchExtractFilesSync,
+  classifyError,
   clearDocumentExtractors,
   clearOcrBackends,
   clearPostProcessors,
@@ -897,6 +914,8 @@ const __version__ = "4.0.0-rc.14";
   extractFile,
   extractFileSync,
   getEmbeddingPreset,
+  getErrorCodeDescription,
+  getErrorCodeName,
   getExtensionsForMime,
   getLastErrorCode,
   getLastPanicContext,