@kreuzberg/wasm 4.0.0-rc.21 → 4.0.0-rc.23

package/dist/index.d.cts

@@ -1,10 +1,226 @@
-import { E as ExtractionConfig, a as ExtractionResult } from './types-CKjcIYcX.cjs.js';
-export { C as Chunk, b as ChunkingConfig, c as ChunkMetadata, d as ExtractedImage, I as ImageExtractionConfig, L as LanguageDetectionConfig, M as Metadata, O as OcrBackendProtocol, e as OcrConfig, P as PageContent, f as PageExtractionConfig, g as PdfConfig, h as PostProcessorConfig, T as Table, i as TesseractConfig, j as TokenReductionConfig, E as ExtractionConfig, a as ExtractionResult } from './types-CKjcIYcX.cjs.js';
+import { E as ExtractionConfig, a as ExtractionResult } from './types-wVLLDHkl.cjs.js';
+export { C as Chunk, b as ChunkingConfig, c as ChunkMetadata, d as ExtractedImage, I as ImageExtractionConfig, L as LanguageDetectionConfig, M as Metadata, O as OcrBackendProtocol, e as OcrConfig, P as PageContent, f as PageExtractionConfig, g as PdfConfig, h as PostProcessorConfig, T as Table, i as TesseractConfig, j as TokenReductionConfig, E as ExtractionConfig, a as ExtractionResult } from './types-wVLLDHkl.cjs.js';
 export { configToJS, fileToUint8Array, isValidExtractionResult, jsToExtractionResult, wrapWasmError } from './adapters/wasm-adapter.cjs';
 export { clearOcrBackends, getOcrBackend, listOcrBackends, registerOcrBackend, unregisterOcrBackend } from './ocr/registry.cjs';
 export { TesseractWasmBackend } from './ocr/tesseract-wasm-backend.cjs';
 export { type RuntimeType, type WasmCapabilities, detectRuntime, getRuntimeInfo, getRuntimeVersion, getWasmCapabilities, hasBigInt, hasBlob, hasFileApi, hasModuleWorkers, hasSharedArrayBuffer, hasWasm, hasWasmStreaming, hasWorkers, isBrowser, isBun, isDeno, isNode, isServerEnvironment, isWebEnvironment } from './runtime.d.cts';
 
+/**
+ * Plugin Registry Module
+ *
+ * This module manages registrations and execution of post-processors and validators
+ * for document extraction pipelines.
+ *
+ * # Thread Safety
+ * All registrations are stored in Maps and are single-threaded safe for WASM environments.
+ *
+ * # Global Callback Functions
+ * The WASM module can invoke processing via global callback functions:
+ * - `__kreuzberg_execute_post_processor`: Execute a registered post-processor
+ * - `__kreuzberg_execute_validator`: Execute a registered validator
+ */
+
+/**
+ * Post-processor plugin interface
+ *
+ * A post-processor modifies extraction results after extraction completes.
+ */
+interface PostProcessor {
+    /**
+     * Get the processor name (must be non-empty string)
+     */
+    name(): string;
+    /**
+     * Get the processing stage (optional, defaults to "middle")
+     * - "early": Process early in the pipeline
+     * - "middle": Process in the middle of the pipeline
+     * - "late": Process late in the pipeline
+     */
+    stage?(): "early" | "middle" | "late";
+    /**
+     * Process an extraction result
+     * Can be sync or async
+     */
+    process(result: ExtractionResult): ExtractionResult | Promise<ExtractionResult>;
+    /**
+     * Shutdown the processor (optional)
+     */
+    shutdown?(): void | Promise<void>;
+}
+/**
+ * Validator plugin interface
+ *
+ * A validator checks extraction results for correctness
+ */
+interface Validator {
+    /**
+     * Get the validator name (must be non-empty string)
+     */
+    name(): string;
+    /**
+     * Get the validation priority (optional, defaults to 50)
+     * Higher numbers = higher priority (execute first)
+     */
+    priority?(): number;
+    /**
+     * Validate an extraction result
+     * Can be sync or async
+     */
+    validate(result: ExtractionResult): {
+        valid: boolean;
+        errors: string[];
+    } | Promise<{
+        valid: boolean;
+        errors: string[];
+    }>;
+    /**
+     * Shutdown the validator (optional)
+     */
+    shutdown?(): void | Promise<void>;
+}
+/**
+ * Register a post-processor plugin
+ *
+ * @param processor - The post-processor to register
+ * @throws {Error} If the processor is invalid or missing required methods
+ *
+ * @example
+ * ```typescript
+ * const processor = {
+ *   name: () => "my-processor",
+ *   stage: () => "middle",
+ *   process: async (result) => {
+ *     result.content = result.content.toUpperCase();
+ *     return result;
+ *   }
+ * };
+ * registerPostProcessor(processor);
+ * ```
+ */
+declare function registerPostProcessor(processor: PostProcessor): void;
+/**
+ * Get a registered post-processor by name
+ *
+ * @param name - The processor name
+ * @returns The processor, or undefined if not found
+ *
+ * @example
+ * ```typescript
+ * const processor = getPostProcessor("my-processor");
+ * if (processor) {
+ *   console.log("Found processor:", processor.name());
+ * }
+ * ```
+ */
+declare function getPostProcessor(name: string): PostProcessor | undefined;
+/**
+ * List all registered post-processor names
+ *
+ * @returns Array of processor names
+ *
+ * @example
+ * ```typescript
+ * const names = listPostProcessors();
+ * console.log("Registered processors:", names);
+ * ```
+ */
+declare function listPostProcessors(): string[];
+/**
+ * Unregister a post-processor and call its shutdown method
+ *
+ * @param name - The processor name
+ * @throws {Error} If the processor is not registered
+ *
+ * @example
+ * ```typescript
+ * await unregisterPostProcessor("my-processor");
+ * ```
+ */
+declare function unregisterPostProcessor(name: string): Promise<void>;
+/**
+ * Clear all registered post-processors
+ *
+ * Calls shutdown on all processors before clearing.
+ *
+ * @example
+ * ```typescript
+ * await clearPostProcessors();
+ * ```
+ */
+declare function clearPostProcessors(): Promise<void>;
+/**
+ * Register a validator plugin
+ *
+ * @param validator - The validator to register
+ * @throws {Error} If the validator is invalid or missing required methods
+ *
+ * @example
+ * ```typescript
+ * const validator = {
+ *   name: () => "my-validator",
+ *   priority: () => 50,
+ *   validate: async (result) => {
+ *     if (!result.content) {
+ *       return { valid: false, errors: ["Content is empty"] };
+ *     }
+ *     return { valid: true, errors: [] };
+ *   }
+ * };
+ * registerValidator(validator);
+ * ```
+ */
+declare function registerValidator(validator: Validator): void;
+/**
+ * Get a registered validator by name
+ *
+ * @param name - The validator name
+ * @returns The validator, or undefined if not found
+ *
+ * @example
+ * ```typescript
+ * const validator = getValidator("my-validator");
+ * if (validator) {
+ *   console.log("Found validator:", validator.name());
+ * }
+ * ```
+ */
+declare function getValidator(name: string): Validator | undefined;
+/**
+ * List all registered validator names
+ *
+ * @returns Array of validator names
+ *
+ * @example
+ * ```typescript
+ * const names = listValidators();
+ * console.log("Registered validators:", names);
+ * ```
+ */
+declare function listValidators(): string[];
+/**
+ * Unregister a validator and call its shutdown method
+ *
+ * @param name - The validator name
+ * @throws {Error} If the validator is not registered
+ *
+ * @example
+ * ```typescript
+ * await unregisterValidator("my-validator");
+ * ```
+ */
+declare function unregisterValidator(name: string): Promise<void>;
+/**
+ * Clear all registered validators
+ *
+ * Calls shutdown on all validators before clearing.
+ *
+ * @example
+ * ```typescript
+ * await clearValidators();
+ * ```
+ */
+declare function clearValidators(): Promise<void>;
+
 /**
  * Kreuzberg - WebAssembly Bindings for Browser and Runtime Environments
  *
@@ -420,4 +636,4 @@ declare function batchExtractFiles(files: File[], config?: ExtractionConfig | nu
  */
 declare function enableOcr(): Promise<void>;
 
-export { batchExtractBytes, batchExtractBytesSync, batchExtractFiles, enableOcr, extractBytes, extractBytesSync, extractFile, extractFromFile, getInitializationError, getVersion, initWasm, isInitialized };
+export { type PostProcessor, type Validator, batchExtractBytes, batchExtractBytesSync, batchExtractFiles, clearPostProcessors, clearValidators, enableOcr, extractBytes, extractBytesSync, extractFile, extractFromFile, getInitializationError, getPostProcessor, getValidator, getVersion, initWasm, isInitialized, listPostProcessors, listValidators, registerPostProcessor, registerValidator, unregisterPostProcessor, unregisterValidator };

package/dist/index.d.ts

@@ -1,10 +1,226 @@
-import { E as ExtractionConfig, a as ExtractionResult } from './types-CKjcIYcX.d.ts';
-export { C as Chunk, b as ChunkingConfig, c as ChunkMetadata, d as ExtractedImage, I as ImageExtractionConfig, L as LanguageDetectionConfig, M as Metadata, O as OcrBackendProtocol, e as OcrConfig, P as PageContent, f as PageExtractionConfig, g as PdfConfig, h as PostProcessorConfig, T as Table, i as TesseractConfig, j as TokenReductionConfig, E as ExtractionConfig, a as ExtractionResult } from './types-CKjcIYcX.d.ts';
+import { E as ExtractionConfig, a as ExtractionResult } from './types-wVLLDHkl.d.ts';
+export { C as Chunk, b as ChunkingConfig, c as ChunkMetadata, d as ExtractedImage, I as ImageExtractionConfig, L as LanguageDetectionConfig, M as Metadata, O as OcrBackendProtocol, e as OcrConfig, P as PageContent, f as PageExtractionConfig, g as PdfConfig, h as PostProcessorConfig, T as Table, i as TesseractConfig, j as TokenReductionConfig, E as ExtractionConfig, a as ExtractionResult } from './types-wVLLDHkl.d.ts';
 export { configToJS, fileToUint8Array, isValidExtractionResult, jsToExtractionResult, wrapWasmError } from './adapters/wasm-adapter.js';
 export { clearOcrBackends, getOcrBackend, listOcrBackends, registerOcrBackend, unregisterOcrBackend } from './ocr/registry.js';
 export { TesseractWasmBackend } from './ocr/tesseract-wasm-backend.js';
 export { type RuntimeType, type WasmCapabilities, detectRuntime, getRuntimeInfo, getRuntimeVersion, getWasmCapabilities, hasBigInt, hasBlob, hasFileApi, hasModuleWorkers, hasSharedArrayBuffer, hasWasm, hasWasmStreaming, hasWorkers, isBrowser, isBun, isDeno, isNode, isServerEnvironment, isWebEnvironment } from './runtime.d.ts';
 
+/**
+ * Plugin Registry Module
+ *
+ * This module manages registrations and execution of post-processors and validators
+ * for document extraction pipelines.
+ *
+ * # Thread Safety
+ * All registrations are stored in Maps and are single-threaded safe for WASM environments.
+ *
+ * # Global Callback Functions
+ * The WASM module can invoke processing via global callback functions:
+ * - `__kreuzberg_execute_post_processor`: Execute a registered post-processor
+ * - `__kreuzberg_execute_validator`: Execute a registered validator
+ */
+
+/**
+ * Post-processor plugin interface
+ *
+ * A post-processor modifies extraction results after extraction completes.
+ */
+interface PostProcessor {
+    /**
+     * Get the processor name (must be non-empty string)
+     */
+    name(): string;
+    /**
+     * Get the processing stage (optional, defaults to "middle")
+     * - "early": Process early in the pipeline
+     * - "middle": Process in the middle of the pipeline
+     * - "late": Process late in the pipeline
+     */
+    stage?(): "early" | "middle" | "late";
+    /**
+     * Process an extraction result
+     * Can be sync or async
+     */
+    process(result: ExtractionResult): ExtractionResult | Promise<ExtractionResult>;
+    /**
+     * Shutdown the processor (optional)
+     */
+    shutdown?(): void | Promise<void>;
+}
+/**
+ * Validator plugin interface
+ *
+ * A validator checks extraction results for correctness
+ */
+interface Validator {
+    /**
+     * Get the validator name (must be non-empty string)
+     */
+    name(): string;
+    /**
+     * Get the validation priority (optional, defaults to 50)
+     * Higher numbers = higher priority (execute first)
+     */
+    priority?(): number;
+    /**
+     * Validate an extraction result
+     * Can be sync or async
+     */
+    validate(result: ExtractionResult): {
+        valid: boolean;
+        errors: string[];
+    } | Promise<{
+        valid: boolean;
+        errors: string[];
+    }>;
+    /**
+     * Shutdown the validator (optional)
+     */
+    shutdown?(): void | Promise<void>;
+}
+/**
+ * Register a post-processor plugin
+ *
+ * @param processor - The post-processor to register
+ * @throws {Error} If the processor is invalid or missing required methods
+ *
+ * @example
+ * ```typescript
+ * const processor = {
+ *   name: () => "my-processor",
+ *   stage: () => "middle",
+ *   process: async (result) => {
+ *     result.content = result.content.toUpperCase();
+ *     return result;
+ *   }
+ * };
+ * registerPostProcessor(processor);
+ * ```
+ */
+declare function registerPostProcessor(processor: PostProcessor): void;
+/**
+ * Get a registered post-processor by name
+ *
+ * @param name - The processor name
+ * @returns The processor, or undefined if not found
+ *
+ * @example
+ * ```typescript
+ * const processor = getPostProcessor("my-processor");
+ * if (processor) {
+ *   console.log("Found processor:", processor.name());
+ * }
+ * ```
+ */
+declare function getPostProcessor(name: string): PostProcessor | undefined;
+/**
+ * List all registered post-processor names
+ *
+ * @returns Array of processor names
+ *
+ * @example
+ * ```typescript
+ * const names = listPostProcessors();
+ * console.log("Registered processors:", names);
+ * ```
+ */
+declare function listPostProcessors(): string[];
+/**
+ * Unregister a post-processor and call its shutdown method
+ *
+ * @param name - The processor name
+ * @throws {Error} If the processor is not registered
+ *
+ * @example
+ * ```typescript
+ * await unregisterPostProcessor("my-processor");
+ * ```
+ */
+declare function unregisterPostProcessor(name: string): Promise<void>;
+/**
+ * Clear all registered post-processors
+ *
+ * Calls shutdown on all processors before clearing.
+ *
+ * @example
+ * ```typescript
+ * await clearPostProcessors();
+ * ```
+ */
+declare function clearPostProcessors(): Promise<void>;
+/**
+ * Register a validator plugin
+ *
+ * @param validator - The validator to register
+ * @throws {Error} If the validator is invalid or missing required methods
+ *
+ * @example
+ * ```typescript
+ * const validator = {
+ *   name: () => "my-validator",
+ *   priority: () => 50,
+ *   validate: async (result) => {
+ *     if (!result.content) {
+ *       return { valid: false, errors: ["Content is empty"] };
+ *     }
+ *     return { valid: true, errors: [] };
+ *   }
+ * };
+ * registerValidator(validator);
+ * ```
+ */
+declare function registerValidator(validator: Validator): void;
+/**
+ * Get a registered validator by name
+ *
+ * @param name - The validator name
+ * @returns The validator, or undefined if not found
+ *
+ * @example
+ * ```typescript
+ * const validator = getValidator("my-validator");
+ * if (validator) {
+ *   console.log("Found validator:", validator.name());
+ * }
+ * ```
+ */
+declare function getValidator(name: string): Validator | undefined;
+/**
+ * List all registered validator names
+ *
+ * @returns Array of validator names
+ *
+ * @example
+ * ```typescript
+ * const names = listValidators();
+ * console.log("Registered validators:", names);
+ * ```
+ */
+declare function listValidators(): string[];
+/**
+ * Unregister a validator and call its shutdown method
+ *
+ * @param name - The validator name
+ * @throws {Error} If the validator is not registered
+ *
+ * @example
+ * ```typescript
+ * await unregisterValidator("my-validator");
+ * ```
+ */
+declare function unregisterValidator(name: string): Promise<void>;
+/**
+ * Clear all registered validators
+ *
+ * Calls shutdown on all validators before clearing.
+ *
+ * @example
+ * ```typescript
+ * await clearValidators();
+ * ```
+ */
+declare function clearValidators(): Promise<void>;
+
 /**
  * Kreuzberg - WebAssembly Bindings for Browser and Runtime Environments
  *
@@ -420,4 +636,4 @@ declare function batchExtractFiles(files: File[], config?: ExtractionConfig | nu
  */
 declare function enableOcr(): Promise<void>;
 
-export { batchExtractBytes, batchExtractBytesSync, batchExtractFiles, enableOcr, extractBytes, extractBytesSync, extractFile, extractFromFile, getInitializationError, getVersion, initWasm, isInitialized };
+export { type PostProcessor, type Validator, batchExtractBytes, batchExtractBytesSync, batchExtractFiles, clearPostProcessors, clearValidators, enableOcr, extractBytes, extractBytesSync, extractFile, extractFromFile, getInitializationError, getPostProcessor, getValidator, getVersion, initWasm, isInitialized, listPostProcessors, listValidators, registerPostProcessor, registerValidator, unregisterPostProcessor, unregisterValidator };