@kreuzberg/wasm 4.0.0-rc.6

package/dist/ocr/tesseract-wasm-backend.mjs

@@ -0,0 +1,424 @@
+class TesseractWasmBackend {
+  /** Tesseract WASM client instance */
+  client = null;
+  /** Track which models are currently loaded to avoid redundant loads */
+  loadedLanguages = /* @__PURE__ */ new Set();
+  /** Cache for language availability validation */
+  supportedLangsCache = null;
+  /** Progress callback for UI updates */
+  progressCallback = null;
+  /** Base URL for training data CDN */
+  CDN_BASE_URL = "https://cdn.jsdelivr.net/npm/tesseract-wasm@0.11.0/dist";
+  /**
+   * Return the unique name of this OCR backend
+   *
+   * @returns Backend identifier "tesseract-wasm"
+   */
+  name() {
+    return "tesseract-wasm";
+  }
+  /**
+   * Return list of supported language codes
+   *
+   * Returns a curated list of commonly available Tesseract language models.
+   * Tesseract supports many more languages through custom models.
+   *
+   * @returns Array of ISO 639-1/2/3 language codes
+   */
+  supportedLanguages() {
+    if (this.supportedLangsCache) {
+      return this.supportedLangsCache;
+    }
+    this.supportedLangsCache = [
+      // Major languages
+      "eng",
+      // English
+      "deu",
+      // German
+      "fra",
+      // French
+      "spa",
+      // Spanish
+      "ita",
+      // Italian
+      "por",
+      // Portuguese
+      "nld",
+      // Dutch
+      "rus",
+      // Russian
+      "jpn",
+      // Japanese
+      "kor",
+      // Korean
+      "chi_sim",
+      // Chinese (Simplified)
+      "chi_tra",
+      // Chinese (Traditional)
+      // Additional European languages
+      "pol",
+      // Polish
+      "tur",
+      // Turkish
+      "swe",
+      // Swedish
+      "dan",
+      // Danish
+      "fin",
+      // Finnish
+      "nor",
+      // Norwegian
+      "ces",
+      // Czech
+      "slk",
+      // Slovak
+      "ron",
+      // Romanian
+      "hun",
+      // Hungarian
+      "hrv",
+      // Croatian
+      "srp",
+      // Serbian
+      "bul",
+      // Bulgarian
+      "ukr",
+      // Ukrainian
+      "ell",
+      // Greek
+      // Asian languages
+      "ara",
+      // Arabic
+      "heb",
+      // Hebrew
+      "hin",
+      // Hindi
+      "tha",
+      // Thai
+      "vie",
+      // Vietnamese
+      "mkd",
+      // Macedonian
+      "ben",
+      // Bengali
+      "tam",
+      // Tamil
+      "tel",
+      // Telugu
+      "kan",
+      // Kannada
+      "mal",
+      // Malayalam
+      "mya",
+      // Burmese
+      "khm",
+      // Khmer
+      "lao",
+      // Lao
+      "sin"
+      // Sinhala
+    ];
+    return this.supportedLangsCache;
+  }
+  /**
+   * Initialize the OCR backend
+   *
+   * Creates the Tesseract WASM client instance. This is called once when
+   * the backend is registered with the extraction pipeline.
+   *
+   * The actual model loading happens in processImage() on-demand to avoid
+   * loading all models upfront.
+   *
+   * @throws {Error} If tesseract-wasm is not available or initialization fails
+   *
+   * @example
+   * ```typescript
+   * const backend = new TesseractWasmBackend();
+   * try {
+   *   await backend.initialize();
+   * } catch (error) {
+   *   console.error('Failed to initialize OCR:', error);
+   * }
+   * ```
+   */
+  async initialize() {
+    if (this.client) {
+      return;
+    }
+    try {
+      const tesseractModule = await this.loadTesseractWasm();
+      if (!tesseractModule || typeof tesseractModule.OCRClient !== "function") {
+        throw new Error("tesseract-wasm OCRClient not found. Ensure tesseract-wasm is installed and available.");
+      }
+      this.client = new tesseractModule.OCRClient();
+      this.loadedLanguages.clear();
+    } catch (error) {
+      const message = error instanceof Error ? error.message : String(error);
+      throw new Error(`Failed to initialize TesseractWasmBackend: ${message}`);
+    }
+  }
+  /**
+   * Process image bytes and extract text via OCR
+   *
+   * Handles image loading, model loading, OCR processing, and result formatting.
+   * Automatically loads the language model on first use and caches it for subsequent calls.
+   *
+   * @param imageBytes - Raw image data (Uint8Array) or Base64-encoded string
+   * @param language - ISO 639-2/3 language code (e.g., "eng", "deu")
+   * @returns Promise resolving to OCR result with content and metadata
+   * @throws {Error} If image processing fails, model loading fails, or language is unsupported
+   *
+   * @example
+   * ```typescript
+   * const backend = new TesseractWasmBackend();
+   * await backend.initialize();
+   *
+   * const imageBuffer = fs.readFileSync('scanned.png');
+   * const result = await backend.processImage(
+   *   new Uint8Array(imageBuffer),
+   *   'eng'
+   * );
+   *
+   * console.log(result.content); // Extracted text
+   * console.log(result.metadata.confidence); // OCR confidence score
+   * ```
+   */
+  async processImage(imageBytes, language) {
+    if (!this.client) {
+      throw new Error("TesseractWasmBackend not initialized. Call initialize() first.");
+    }
+    const supported = this.supportedLanguages();
+    const normalizedLang = language.toLowerCase();
+    const isSupported = supported.some((lang) => lang.toLowerCase() === normalizedLang);
+    if (!isSupported) {
+      throw new Error(`Language "${language}" is not supported. Supported languages: ${supported.join(", ")}`);
+    }
+    try {
+      if (!this.loadedLanguages.has(normalizedLang)) {
+        this.reportProgress(10);
+        await this.loadLanguageModel(normalizedLang);
+        this.loadedLanguages.add(normalizedLang);
+        this.reportProgress(30);
+      }
+      this.reportProgress(40);
+      const imageBitmap = await this.convertToImageBitmap(imageBytes);
+      this.reportProgress(50);
+      await this.client.loadImage(imageBitmap);
+      this.reportProgress(70);
+      const text = await this.client.getText();
+      const confidence = await this.getConfidenceScore();
+      const pageMetadata = await this.getPageMetadata();
+      this.reportProgress(90);
+      return {
+        content: text,
+        mime_type: "text/plain",
+        metadata: {
+          language: normalizedLang,
+          confidence,
+          ...pageMetadata
+        },
+        tables: []
+        // Tesseract-wasm doesn't provide structured table detection
+      };
+    } catch (error) {
+      const message = error instanceof Error ? error.message : String(error);
+      throw new Error(`OCR processing failed for language "${language}": ${message}`);
+    } finally {
+      this.reportProgress(100);
+    }
+  }
+  /**
+   * Shutdown the OCR backend and release resources
+   *
+   * Properly cleans up the Tesseract WASM client, freeing memory and Web Workers.
+   * Called when the backend is unregistered or the application shuts down.
+   *
+   * @throws {Error} If cleanup fails (errors are logged but not critical)
+   *
+   * @example
+   * ```typescript
+   * const backend = new TesseractWasmBackend();
+   * await backend.initialize();
+   * // ... use backend ...
+   * await backend.shutdown(); // Clean up resources
+   * ```
+   */
+  async shutdown() {
+    try {
+      if (this.client) {
+        if (typeof this.client.destroy === "function") {
+          this.client.destroy();
+        }
+        if (typeof this.client.terminate === "function") {
+          this.client.terminate();
+        }
+        this.client = null;
+      }
+      this.loadedLanguages.clear();
+      this.supportedLangsCache = null;
+      this.progressCallback = null;
+    } catch (error) {
+      console.warn(
+        `Warning during TesseractWasmBackend shutdown: ${error instanceof Error ? error.message : String(error)}`
+      );
+    }
+  }
+  /**
+   * Set a progress callback for UI updates
+   *
+   * Allows the UI to display progress during OCR processing.
+   * The callback will be called with values from 0 to 100.
+   *
+   * @param callback - Function to call with progress percentage
+   *
+   * @example
+   * ```typescript
+   * const backend = new TesseractWasmBackend();
+   * backend.setProgressCallback((progress) => {
+   *   console.log(`OCR Progress: ${progress}%`);
+   *   document.getElementById('progress-bar').style.width = `${progress}%`;
+   * });
+   * ```
+   */
+  setProgressCallback(callback) {
+    this.progressCallback = callback;
+  }
+  /**
+   * Load language model from CDN
+   *
+   * Fetches the training data for a specific language from jsDelivr CDN.
+   * This is an MVP approach - models are cached by the browser.
+   *
+   * @param language - ISO 639-2/3 language code
+   * @throws {Error} If model download fails or language is not available
+   *
+   * @internal
+   */
+  async loadLanguageModel(language) {
+    if (!this.client) {
+      throw new Error("Client not initialized");
+    }
+    const modelFilename = `${language}.traineddata`;
+    const modelUrl = `${this.CDN_BASE_URL}/${modelFilename}`;
+    try {
+      await this.client.loadModel(modelUrl);
+    } catch (error) {
+      const message = error instanceof Error ? error.message : String(error);
+      throw new Error(`Failed to load model for language "${language}" from ${modelUrl}: ${message}`);
+    }
+  }
+  /**
+   * Convert image bytes or Base64 string to ImageBitmap
+   *
+   * Handles both Uint8Array and Base64-encoded image data, converting to
+   * ImageBitmap format required by Tesseract WASM.
+   *
+   * @param imageBytes - Image data as Uint8Array or Base64 string
+   * @returns Promise resolving to ImageBitmap
+   * @throws {Error} If conversion fails (browser API not available or invalid image data)
+   *
+   * @internal
+   */
+  async convertToImageBitmap(imageBytes) {
+    if (typeof createImageBitmap === "undefined") {
+      throw new Error("createImageBitmap is not available. TesseractWasmBackend requires a browser environment.");
+    }
+    try {
+      let bytes = imageBytes;
+      if (typeof imageBytes === "string") {
+        const binaryString = atob(imageBytes);
+        bytes = new Uint8Array(binaryString.length);
+        for (let i = 0; i < binaryString.length; i++) {
+          bytes[i] = binaryString.charCodeAt(i);
+        }
+      }
+      const blob = new Blob([bytes]);
+      const imageBitmap = await createImageBitmap(blob);
+      return imageBitmap;
+    } catch (error) {
+      const message = error instanceof Error ? error.message : String(error);
+      throw new Error(`Failed to convert image bytes to ImageBitmap: ${message}`);
+    }
+  }
+  /**
+   * Get confidence score from OCR result
+   *
+   * Attempts to retrieve confidence score from Tesseract.
+   * Returns a safe default if unavailable.
+   *
+   * @returns Confidence score between 0 and 1
+   *
+   * @internal
+   */
+  async getConfidenceScore() {
+    try {
+      if (this.client && typeof this.client.getConfidence === "function") {
+        const confidence = await this.client.getConfidence();
+        return confidence > 1 ? confidence / 100 : confidence;
+      }
+    } catch {
+    }
+    return 0.9;
+  }
+  /**
+   * Get page metadata from OCR result
+   *
+   * Retrieves additional metadata like image dimensions and processing info.
+   *
+   * @returns Metadata object (may be empty if unavailable)
+   *
+   * @internal
+   */
+  async getPageMetadata() {
+    try {
+      if (this.client && typeof this.client.getPageMetadata === "function") {
+        return await this.client.getPageMetadata();
+      }
+    } catch {
+    }
+    return {};
+  }
+  /**
+   * Dynamically load tesseract-wasm module
+   *
+   * Uses dynamic import to load tesseract-wasm only when needed,
+   * avoiding hard dependency in browser environments where it may not be bundled.
+   *
+   * @returns tesseract-wasm module object
+   * @throws {Error} If module cannot be imported
+   *
+   * @internal
+   */
+  async loadTesseractWasm() {
+    try {
+      const module = await import("tesseract-wasm");
+      return module;
+    } catch (error) {
+      const message = error instanceof Error ? error.message : String(error);
+      throw new Error(
+        `Failed to import tesseract-wasm. Ensure it is installed via: npm install tesseract-wasm. Error: ${message}`
+      );
+    }
+  }
+  /**
+   * Report progress to progress callback
+   *
+   * Internal helper for notifying progress updates during OCR processing.
+   *
+   * @param progress - Progress percentage (0-100)
+   *
+   * @internal
+   */
+  reportProgress(progress) {
+    if (this.progressCallback) {
+      try {
+        this.progressCallback(Math.min(100, Math.max(0, progress)));
+      } catch {
+      }
+    }
+  }
+}
+export {
+  TesseractWasmBackend
+};
+//# sourceMappingURL=tesseract-wasm-backend.mjs.map

package/dist/ocr/tesseract-wasm-backend.mjs.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../typescript/ocr/tesseract-wasm-backend.ts"],"sourcesContent":["/**\n * Tesseract WASM OCR Backend\n *\n * Provides OCR capabilities using tesseract-wasm library for browser environments.\n * Loads training data on-demand from jsDelivr CDN and implements the OcrBackendProtocol.\n *\n * ## Browser-Only Requirement\n *\n * This backend requires browser APIs like createImageBitmap and Web Workers.\n * It will NOT work in Node.js environments without additional canvas polyfills.\n *\n * ## Supported Languages\n *\n * Common ISO 639-1 and ISO 639-2 codes:\n * - English: \"eng\"\n * - German: \"deu\"\n * - French: \"fra\"\n * - Spanish: \"spa\"\n * - Italian: \"ita\"\n * - Portuguese: \"por\"\n * - Dutch: \"nld\"\n * - Russian: \"rus\"\n * - Chinese (Simplified): \"chi_sim\"\n * - Chinese (Traditional): \"chi_tra\"\n * - Japanese: \"jpn\"\n * - Korean: \"kor\"\n * - Arabic: \"ara\"\n * - Hindi: \"hin\"\n *\n * For complete language list, see: https://github.com/naptha/tesseract.js\n *\n * @example Basic Usage\n * ```typescript\n * import { TesseractWasmBackend } from '@kreuzberg/wasm/ocr/tesseract-wasm-backend';\n * import { registerOcrBackend, extractBytes, initWasm } from '@kreuzberg/wasm';\n *\n * // Initialize\n * await initWasm();\n * const backend = new TesseractWasmBackend();\n * await backend.initialize();\n * registerOcrBackend(backend);\n *\n * // Use in extraction\n * const imageBytes = new Uint8Array(buffer);\n * const result = await extractBytes(imageBytes, 'image/png', {\n *   ocr: { backend: 'tesseract-wasm', language: 'eng' }\n * });\n * console.log(result.content); // Extracted text\n * ```\n *\n * @example With Language Auto-Detection\n * ```typescript\n * const backend = new TesseractWasmBackend();\n * await backend.initialize();\n * registerOcrBackend(backend);\n *\n * // Extract without specifying language - backend will auto-detect\n * const result = await extractBytes(imageBytes, 'image/png', {\n *   ocr: { backend: 'tesseract-wasm' } // language will auto-detect\n * });\n * ```\n */\n\nimport type { OcrBackendProtocol } from \"../types.js\";\n\n/**\n * Tesseract WASM Client interface\n * Type definition for tesseract-wasm's OCRClient class\n */\ninterface TesseractClient {\n\tloadModel(modelPath: string): Promise<void>;\n\tloadImage(image: ImageBitmap | Blob): Promise<void>;\n\tgetText(): Promise<string>;\n\tgetConfidence(): Promise<number>;\n\tgetPageMetadata(): Promise<Record<string, unknown>>;\n\tdestroy(): void;\n\tterminate(): void;\n}\n\n/**\n * TesseractWasmBackend - OCR backend using tesseract-wasm library\n *\n * Implements the OcrBackendProtocol for Kreuzberg document extraction pipeline.\n * Provides comprehensive OCR support with model caching, error handling, and progress reporting.\n */\nexport class TesseractWasmBackend implements OcrBackendProtocol {\n\t/** Tesseract WASM client instance */\n\tprivate client: TesseractClient | null = null;\n\n\t/** Track which models are currently loaded to avoid redundant loads */\n\tprivate loadedLanguages: Set<string> = new Set();\n\n\t/** Cache for language availability validation */\n\tprivate supportedLangsCache: string[] | null = null;\n\n\t/** Progress callback for UI updates */\n\tprivate progressCallback: ((progress: number) => void) | null = null;\n\n\t/** Base URL for training data CDN */\n\tprivate readonly CDN_BASE_URL = \"https://cdn.jsdelivr.net/npm/tesseract-wasm@0.11.0/dist\";\n\n\t/**\n\t * Return the unique name of this OCR backend\n\t *\n\t * @returns Backend identifier \"tesseract-wasm\"\n\t */\n\tname(): string {\n\t\treturn \"tesseract-wasm\";\n\t}\n\n\t/**\n\t * Return list of supported language codes\n\t *\n\t * Returns a curated list of commonly available Tesseract language models.\n\t * Tesseract supports many more languages through custom models.\n\t *\n\t * @returns Array of ISO 639-1/2/3 language codes\n\t */\n\tsupportedLanguages(): string[] {\n\t\t// Return cached list if already computed\n\t\tif (this.supportedLangsCache) {\n\t\t\treturn this.supportedLangsCache;\n\t\t}\n\n\t\t// Comprehensive list of languages supported by tesseract-wasm\n\t\t// Includes both 3-letter (ISO 639-2) and 2-letter (ISO 639-1) codes where applicable\n\t\tthis.supportedLangsCache = [\n\t\t\t// Major languages\n\t\t\t\"eng\", // English\n\t\t\t\"deu\", // German\n\t\t\t\"fra\", // French\n\t\t\t\"spa\", // Spanish\n\t\t\t\"ita\", // Italian\n\t\t\t\"por\", // Portuguese\n\t\t\t\"nld\", // Dutch\n\t\t\t\"rus\", // Russian\n\t\t\t\"jpn\", // Japanese\n\t\t\t\"kor\", // Korean\n\t\t\t\"chi_sim\", // Chinese (Simplified)\n\t\t\t\"chi_tra\", // Chinese (Traditional)\n\n\t\t\t// Additional European languages\n\t\t\t\"pol\", // Polish\n\t\t\t\"tur\", // Turkish\n\t\t\t\"swe\", // Swedish\n\t\t\t\"dan\", // Danish\n\t\t\t\"fin\", // Finnish\n\t\t\t\"nor\", // Norwegian\n\t\t\t\"ces\", // Czech\n\t\t\t\"slk\", // Slovak\n\t\t\t\"ron\", // Romanian\n\t\t\t\"hun\", // Hungarian\n\t\t\t\"hrv\", // Croatian\n\t\t\t\"srp\", // Serbian\n\t\t\t\"bul\", // Bulgarian\n\t\t\t\"ukr\", // Ukrainian\n\t\t\t\"ell\", // Greek\n\n\t\t\t// Asian languages\n\t\t\t\"ara\", // Arabic\n\t\t\t\"heb\", // Hebrew\n\t\t\t\"hin\", // Hindi\n\t\t\t\"tha\", // Thai\n\t\t\t\"vie\", // Vietnamese\n\t\t\t\"mkd\", // Macedonian\n\t\t\t\"ben\", // Bengali\n\t\t\t\"tam\", // Tamil\n\t\t\t\"tel\", // Telugu\n\t\t\t\"kan\", // Kannada\n\t\t\t\"mal\", // Malayalam\n\t\t\t\"mya\", // Burmese\n\t\t\t\"khm\", // Khmer\n\t\t\t\"lao\", // Lao\n\t\t\t\"sin\", // Sinhala\n\t\t];\n\n\t\treturn this.supportedLangsCache;\n\t}\n\n\t/**\n\t * Initialize the OCR backend\n\t *\n\t * Creates the Tesseract WASM client instance. This is called once when\n\t * the backend is registered with the extraction pipeline.\n\t *\n\t * The actual model loading happens in processImage() on-demand to avoid\n\t * loading all models upfront.\n\t *\n\t * @throws {Error} If tesseract-wasm is not available or initialization fails\n\t *\n\t * @example\n\t * ```typescript\n\t * const backend = new TesseractWasmBackend();\n\t * try {\n\t *   await backend.initialize();\n\t * } catch (error) {\n\t *   console.error('Failed to initialize OCR:', error);\n\t * }\n\t * ```\n\t */\n\tasync initialize(): Promise<void> {\n\t\tif (this.client) {\n\t\t\treturn; // Already initialized\n\t\t}\n\n\t\ttry {\n\t\t\t// Dynamically import tesseract-wasm\n\t\t\tconst tesseractModule = await this.loadTesseractWasm();\n\n\t\t\t// @ts-expect-error - tesseract-wasm types are not fully typed\n\t\t\tif (!tesseractModule || typeof tesseractModule.OCRClient !== \"function\") {\n\t\t\t\tthrow new Error(\"tesseract-wasm OCRClient not found. Ensure tesseract-wasm is installed and available.\");\n\t\t\t}\n\n\t\t\t// Create client instance\n\t\t\t// @ts-expect-error - tesseract-wasm types are not fully typed\n\t\t\tthis.client = new tesseractModule.OCRClient();\n\n\t\t\t// Initialize tracking\n\t\t\tthis.loadedLanguages.clear();\n\t\t} catch (error) {\n\t\t\tconst message = error instanceof Error ? error.message : String(error);\n\t\t\tthrow new Error(`Failed to initialize TesseractWasmBackend: ${message}`);\n\t\t}\n\t}\n\n\t/**\n\t * Process image bytes and extract text via OCR\n\t *\n\t * Handles image loading, model loading, OCR processing, and result formatting.\n\t * Automatically loads the language model on first use and caches it for subsequent calls.\n\t *\n\t * @param imageBytes - Raw image data (Uint8Array) or Base64-encoded string\n\t * @param language - ISO 639-2/3 language code (e.g., \"eng\", \"deu\")\n\t * @returns Promise resolving to OCR result with content and metadata\n\t * @throws {Error} If image processing fails, model loading fails, or language is unsupported\n\t *\n\t * @example\n\t * ```typescript\n\t * const backend = new TesseractWasmBackend();\n\t * await backend.initialize();\n\t *\n\t * const imageBuffer = fs.readFileSync('scanned.png');\n\t * const result = await backend.processImage(\n\t *   new Uint8Array(imageBuffer),\n\t *   'eng'\n\t * );\n\t *\n\t * console.log(result.content); // Extracted text\n\t * console.log(result.metadata.confidence); // OCR confidence score\n\t * ```\n\t */\n\tasync processImage(\n\t\timageBytes: Uint8Array | string,\n\t\tlanguage: string,\n\t): Promise<{\n\t\tcontent: string;\n\t\tmime_type: string;\n\t\tmetadata: Record<string, unknown>;\n\t\ttables: unknown[];\n\t}> {\n\t\tif (!this.client) {\n\t\t\tthrow new Error(\"TesseractWasmBackend not initialized. Call initialize() first.\");\n\t\t}\n\n\t\t// Validate language support\n\t\tconst supported = this.supportedLanguages();\n\t\t// Normalize language code for comparison\n\t\tconst normalizedLang = language.toLowerCase();\n\t\tconst isSupported = supported.some((lang) => lang.toLowerCase() === normalizedLang);\n\n\t\tif (!isSupported) {\n\t\t\tthrow new Error(`Language \"${language}\" is not supported. Supported languages: ${supported.join(\", \")}`);\n\t\t}\n\n\t\ttry {\n\t\t\t// Load language model if not already loaded\n\t\t\tif (!this.loadedLanguages.has(normalizedLang)) {\n\t\t\t\tthis.reportProgress(10); // Progress: loading model\n\t\t\t\tawait this.loadLanguageModel(normalizedLang);\n\t\t\t\tthis.loadedLanguages.add(normalizedLang);\n\t\t\t\tthis.reportProgress(30); // Progress: model loaded\n\t\t\t}\n\n\t\t\t// Convert image bytes to ImageBitmap\n\t\t\tthis.reportProgress(40); // Progress: processing image\n\t\t\tconst imageBitmap = await this.convertToImageBitmap(imageBytes);\n\n\t\t\t// Load image into Tesseract\n\t\t\tthis.reportProgress(50); // Progress: loading image\n\t\t\tawait this.client.loadImage(imageBitmap);\n\n\t\t\t// Perform OCR\n\t\t\tthis.reportProgress(70); // Progress: performing OCR\n\t\t\tconst text = await this.client.getText();\n\n\t\t\t// Get confidence and metadata\n\t\t\tconst confidence = await this.getConfidenceScore();\n\t\t\tconst pageMetadata = await this.getPageMetadata();\n\n\t\t\tthis.reportProgress(90); // Progress: nearly complete\n\n\t\t\t// Return result in Kreuzberg format\n\t\t\treturn {\n\t\t\t\tcontent: text,\n\t\t\t\tmime_type: \"text/plain\",\n\t\t\t\tmetadata: {\n\t\t\t\t\tlanguage: normalizedLang,\n\t\t\t\t\tconfidence,\n\t\t\t\t\t...pageMetadata,\n\t\t\t\t},\n\t\t\t\ttables: [], // Tesseract-wasm doesn't provide structured table detection\n\t\t\t};\n\t\t} catch (error) {\n\t\t\tconst message = error instanceof Error ? error.message : String(error);\n\t\t\tthrow new Error(`OCR processing failed for language \"${language}\": ${message}`);\n\t\t} finally {\n\t\t\tthis.reportProgress(100); // Progress: complete\n\t\t}\n\t}\n\n\t/**\n\t * Shutdown the OCR backend and release resources\n\t *\n\t * Properly cleans up the Tesseract WASM client, freeing memory and Web Workers.\n\t * Called when the backend is unregistered or the application shuts down.\n\t *\n\t * @throws {Error} If cleanup fails (errors are logged but not critical)\n\t *\n\t * @example\n\t * ```typescript\n\t * const backend = new TesseractWasmBackend();\n\t * await backend.initialize();\n\t * // ... use backend ...\n\t * await backend.shutdown(); // Clean up resources\n\t * ```\n\t */\n\tasync shutdown(): Promise<void> {\n\t\ttry {\n\t\t\tif (this.client) {\n\t\t\t\t// Try both destroy and terminate for compatibility\n\t\t\t\tif (typeof this.client.destroy === \"function\") {\n\t\t\t\t\tthis.client.destroy();\n\t\t\t\t}\n\t\t\t\tif (typeof this.client.terminate === \"function\") {\n\t\t\t\t\tthis.client.terminate();\n\t\t\t\t}\n\t\t\t\tthis.client = null;\n\t\t\t}\n\n\t\t\t// Clear cached state\n\t\t\tthis.loadedLanguages.clear();\n\t\t\tthis.supportedLangsCache = null;\n\t\t\tthis.progressCallback = null;\n\t\t} catch (error) {\n\t\t\t// Log error but don't throw - shutdown is best-effort\n\t\t\tconsole.warn(\n\t\t\t\t`Warning during TesseractWasmBackend shutdown: ${error instanceof Error ? error.message : String(error)}`,\n\t\t\t);\n\t\t}\n\t}\n\n\t/**\n\t * Set a progress callback for UI updates\n\t *\n\t * Allows the UI to display progress during OCR processing.\n\t * The callback will be called with values from 0 to 100.\n\t *\n\t * @param callback - Function to call with progress percentage\n\t *\n\t * @example\n\t * ```typescript\n\t * const backend = new TesseractWasmBackend();\n\t * backend.setProgressCallback((progress) => {\n\t *   console.log(`OCR Progress: ${progress}%`);\n\t *   document.getElementById('progress-bar').style.width = `${progress}%`;\n\t * });\n\t * ```\n\t */\n\tsetProgressCallback(callback: (progress: number) => void): void {\n\t\tthis.progressCallback = callback;\n\t}\n\n\t/**\n\t * Load language model from CDN\n\t *\n\t * Fetches the training data for a specific language from jsDelivr CDN.\n\t * This is an MVP approach - models are cached by the browser.\n\t *\n\t * @param language - ISO 639-2/3 language code\n\t * @throws {Error} If model download fails or language is not available\n\t *\n\t * @internal\n\t */\n\tprivate async loadLanguageModel(language: string): Promise<void> {\n\t\tif (!this.client) {\n\t\t\tthrow new Error(\"Client not initialized\");\n\t\t}\n\n\t\t// Construct model URL - models are named with their language code\n\t\tconst modelFilename = `${language}.traineddata`;\n\t\tconst modelUrl = `${this.CDN_BASE_URL}/${modelFilename}`;\n\n\t\ttry {\n\t\t\tawait this.client.loadModel(modelUrl);\n\t\t} catch (error) {\n\t\t\tconst message = error instanceof Error ? error.message : String(error);\n\t\t\tthrow new Error(`Failed to load model for language \"${language}\" from ${modelUrl}: ${message}`);\n\t\t}\n\t}\n\n\t/**\n\t * Convert image bytes or Base64 string to ImageBitmap\n\t *\n\t * Handles both Uint8Array and Base64-encoded image data, converting to\n\t * ImageBitmap format required by Tesseract WASM.\n\t *\n\t * @param imageBytes - Image data as Uint8Array or Base64 string\n\t * @returns Promise resolving to ImageBitmap\n\t * @throws {Error} If conversion fails (browser API not available or invalid image data)\n\t *\n\t * @internal\n\t */\n\tprivate async convertToImageBitmap(imageBytes: Uint8Array | string): Promise<ImageBitmap> {\n\t\t// Check if createImageBitmap is available (browser only)\n\t\tif (typeof createImageBitmap === \"undefined\") {\n\t\t\tthrow new Error(\"createImageBitmap is not available. TesseractWasmBackend requires a browser environment.\");\n\t\t}\n\n\t\ttry {\n\t\t\t// Convert to Uint8Array if string (Base64)\n\t\t\tlet bytes = imageBytes;\n\t\t\tif (typeof imageBytes === \"string\") {\n\t\t\t\t// Decode Base64 to binary\n\t\t\t\tconst binaryString = atob(imageBytes);\n\t\t\t\tbytes = new Uint8Array(binaryString.length);\n\t\t\t\tfor (let i = 0; i < binaryString.length; i++) {\n\t\t\t\t\t(bytes as Uint8Array)[i] = binaryString.charCodeAt(i);\n\t\t\t\t}\n\t\t\t}\n\n\t\t\t// Create Blob from bytes\n\t\t\tconst blob = new Blob([bytes as Uint8Array] as BlobPart[]);\n\n\t\t\t// Convert Blob to ImageBitmap\n\t\t\tconst imageBitmap = await createImageBitmap(blob);\n\t\t\treturn imageBitmap;\n\t\t} catch (error) {\n\t\t\tconst message = error instanceof Error ? error.message : String(error);\n\t\t\tthrow new Error(`Failed to convert image bytes to ImageBitmap: ${message}`);\n\t\t}\n\t}\n\n\t/**\n\t * Get confidence score from OCR result\n\t *\n\t * Attempts to retrieve confidence score from Tesseract.\n\t * Returns a safe default if unavailable.\n\t *\n\t * @returns Confidence score between 0 and 1\n\t *\n\t * @internal\n\t */\n\tprivate async getConfidenceScore(): Promise<number> {\n\t\ttry {\n\t\t\tif (this.client && typeof this.client.getConfidence === \"function\") {\n\t\t\t\tconst confidence = await this.client.getConfidence();\n\t\t\t\t// Normalize to 0-1 range if needed (some versions return 0-100)\n\t\t\t\treturn confidence > 1 ? confidence / 100 : confidence;\n\t\t\t}\n\t\t} catch {\n\t\t\t// Silently fail - confidence is optional\n\t\t}\n\t\treturn 0.9; // Default reasonable confidence\n\t}\n\n\t/**\n\t * Get page metadata from OCR result\n\t *\n\t * Retrieves additional metadata like image dimensions and processing info.\n\t *\n\t * @returns Metadata object (may be empty if unavailable)\n\t *\n\t * @internal\n\t */\n\tprivate async getPageMetadata(): Promise<Record<string, unknown>> {\n\t\ttry {\n\t\t\tif (this.client && typeof this.client.getPageMetadata === \"function\") {\n\t\t\t\treturn await this.client.getPageMetadata();\n\t\t\t}\n\t\t} catch {\n\t\t\t// Silently fail - metadata is optional\n\t\t}\n\t\treturn {};\n\t}\n\n\t/**\n\t * Dynamically load tesseract-wasm module\n\t *\n\t * Uses dynamic import to load tesseract-wasm only when needed,\n\t * avoiding hard dependency in browser environments where it may not be bundled.\n\t *\n\t * @returns tesseract-wasm module object\n\t * @throws {Error} If module cannot be imported\n\t *\n\t * @internal\n\t */\n\tprivate async loadTesseractWasm(): Promise<unknown> {\n\t\ttry {\n\t\t\t// Use dynamic import to handle both ESM and CJS\n\t\t\t// @ts-expect-error - tesseract-wasm has package.json exports issues with TypeScript\n\t\t\tconst module = await import(\"tesseract-wasm\");\n\t\t\treturn module;\n\t\t} catch (error) {\n\t\t\tconst message = error instanceof Error ? error.message : String(error);\n\t\t\tthrow new Error(\n\t\t\t\t`Failed to import tesseract-wasm. Ensure it is installed via: npm install tesseract-wasm. Error: ${message}`,\n\t\t\t);\n\t\t}\n\t}\n\n\t/**\n\t * Report progress to progress callback\n\t *\n\t * Internal helper for notifying progress updates during OCR processing.\n\t *\n\t * @param progress - Progress percentage (0-100)\n\t *\n\t * @internal\n\t */\n\tprivate reportProgress(progress: number): void {\n\t\tif (this.progressCallback) {\n\t\t\ttry {\n\t\t\t\tthis.progressCallback(Math.min(100, Math.max(0, progress)));\n\t\t\t} catch {\n\t\t\t\t// Ignore callback errors to prevent blocking OCR processing\n\t\t\t}\n\t\t}\n\t}\n}\n"],"mappings":"AAqFO,MAAM,qBAAmD;AAAA;AAAA,EAEvD,SAAiC;AAAA;AAAA,EAGjC,kBAA+B,oBAAI,IAAI;AAAA;AAAA,EAGvC,sBAAuC;AAAA;AAAA,EAGvC,mBAAwD;AAAA;AAAA,EAG/C,eAAe;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAOhC,OAAe;AACd,WAAO;AAAA,EACR;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAUA,qBAA+B;AAE9B,QAAI,KAAK,qBAAqB;AAC7B,aAAO,KAAK;AAAA,IACb;AAIA,SAAK,sBAAsB;AAAA;AAAA,MAE1B;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA;AAAA,MAGA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA;AAAA,MAGA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,MACA;AAAA;AAAA,IACD;AAEA,WAAO,KAAK;AAAA,EACb;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAuBA,MAAM,aAA4B;AACjC,QAAI,KAAK,QAAQ;AAChB;AAAA,IACD;AAEA,QAAI;AAEH,YAAM,kBAAkB,MAAM,KAAK,kBAAkB;AAGrD,UAAI,CAAC,mBAAmB,OAAO,gBAAgB,cAAc,YAAY;AACxE,cAAM,IAAI,MAAM,uFAAuF;AAAA,MACxG;AAIA,WAAK,SAAS,IAAI,gBAAgB,UAAU;AAG5C,WAAK,gBAAgB,MAAM;AAAA,IAC5B,SAAS,OAAO;AACf,YAAM,UAAU,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK;AACrE,YAAM,IAAI,MAAM,8CAA8C,OAAO,EAAE;AAAA,IACxE;AAAA,EACD;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EA4BA,MAAM,aACL,YACA,UAME;AACF,QAAI,CAAC,KAAK,QAAQ;AACjB,YAAM,IAAI,MAAM,gEAAgE;AAAA,IACjF;AAGA,UAAM,YAAY,KAAK,mBAAmB;AAE1C,UAAM,iBAAiB,SAAS,YAAY;AAC5C,UAAM,cAAc,UAAU,KAAK,CAAC,SAAS,KAAK,YAAY,MAAM,cAAc;AAElF,QAAI,CAAC,aAAa;AACjB,YAAM,IAAI,MAAM,aAAa,QAAQ,4CAA4C,UAAU,KAAK,IAAI,CAAC,EAAE;AAAA,IACxG;AAEA,QAAI;AAEH,UAAI,CAAC,KAAK,gBAAgB,IAAI,cAAc,GAAG;AAC9C,aAAK,eAAe,EAAE;AACtB,cAAM,KAAK,kBAAkB,cAAc;AAC3C,aAAK,gBAAgB,IAAI,cAAc;AACvC,aAAK,eAAe,EAAE;AAAA,MACvB;AAGA,WAAK,eAAe,EAAE;AACtB,YAAM,cAAc,MAAM,KAAK,qBAAqB,UAAU;AAG9D,WAAK,eAAe,EAAE;AACtB,YAAM,KAAK,OAAO,UAAU,WAAW;AAGvC,WAAK,eAAe,EAAE;AACtB,YAAM,OAAO,MAAM,KAAK,OAAO,QAAQ;AAGvC,YAAM,aAAa,MAAM,KAAK,mBAAmB;AACjD,YAAM,eAAe,MAAM,KAAK,gBAAgB;AAEhD,WAAK,eAAe,EAAE;AAGtB,aAAO;AAAA,QACN,SAAS;AAAA,QACT,WAAW;AAAA,QACX,UAAU;AAAA,UACT,UAAU;AAAA,UACV;AAAA,UACA,GAAG;AAAA,QACJ;AAAA,QACA,QAAQ,CAAC;AAAA;AAAA,MACV;AAAA,IACD,SAAS,OAAO;AACf,YAAM,UAAU,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK;AACrE,YAAM,IAAI,MAAM,uCAAuC,QAAQ,MAAM,OAAO,EAAE;AAAA,IAC/E,UAAE;AACD,WAAK,eAAe,GAAG;AAAA,IACxB;AAAA,EACD;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAkBA,MAAM,WAA0B;AAC/B,QAAI;AACH,UAAI,KAAK,QAAQ;AAEhB,YAAI,OAAO,KAAK,OAAO,YAAY,YAAY;AAC9C,eAAK,OAAO,QAAQ;AAAA,QACrB;AACA,YAAI,OAAO,KAAK,OAAO,cAAc,YAAY;AAChD,eAAK,OAAO,UAAU;AAAA,QACvB;AACA,aAAK,SAAS;AAAA,MACf;AAGA,WAAK,gBAAgB,MAAM;AAC3B,WAAK,sBAAsB;AAC3B,WAAK,mBAAmB;AAAA,IACzB,SAAS,OAAO;AAEf,cAAQ;AAAA,QACP,iDAAiD,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK,CAAC;AAAA,MACxG;AAAA,IACD;AAAA,EACD;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAmBA,oBAAoB,UAA4C;AAC/D,SAAK,mBAAmB;AAAA,EACzB;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAaA,MAAc,kBAAkB,UAAiC;AAChE,QAAI,CAAC,KAAK,QAAQ;AACjB,YAAM,IAAI,MAAM,wBAAwB;AAAA,IACzC;AAGA,UAAM,gBAAgB,GAAG,QAAQ;AACjC,UAAM,WAAW,GAAG,KAAK,YAAY,IAAI,aAAa;AAEtD,QAAI;AACH,YAAM,KAAK,OAAO,UAAU,QAAQ;AAAA,IACrC,SAAS,OAAO;AACf,YAAM,UAAU,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK;AACrE,YAAM,IAAI,MAAM,sCAAsC,QAAQ,UAAU,QAAQ,KAAK,OAAO,EAAE;AAAA,IAC/F;AAAA,EACD;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAcA,MAAc,qBAAqB,YAAuD;AAEzF,QAAI,OAAO,sBAAsB,aAAa;AAC7C,YAAM,IAAI,MAAM,0FAA0F;AAAA,IAC3G;AAEA,QAAI;AAEH,UAAI,QAAQ;AACZ,UAAI,OAAO,eAAe,UAAU;AAEnC,cAAM,eAAe,KAAK,UAAU;AACpC,gBAAQ,IAAI,WAAW,aAAa,MAAM;AAC1C,iBAAS,IAAI,GAAG,IAAI,aAAa,QAAQ,KAAK;AAC7C,UAAC,MAAqB,CAAC,IAAI,aAAa,WAAW,CAAC;AAAA,QACrD;AAAA,MACD;AAGA,YAAM,OAAO,IAAI,KAAK,CAAC,KAAmB,CAAe;AAGzD,YAAM,cAAc,MAAM,kBAAkB,IAAI;AAChD,aAAO;AAAA,IACR,SAAS,OAAO;AACf,YAAM,UAAU,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK;AACrE,YAAM,IAAI,MAAM,iDAAiD,OAAO,EAAE;AAAA,IAC3E;AAAA,EACD;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAYA,MAAc,qBAAsC;AACnD,QAAI;AACH,UAAI,KAAK,UAAU,OAAO,KAAK,OAAO,kBAAkB,YAAY;AACnE,cAAM,aAAa,MAAM,KAAK,OAAO,cAAc;AAEnD,eAAO,aAAa,IAAI,aAAa,MAAM;AAAA,MAC5C;AAAA,IACD,QAAQ;AAAA,IAER;AACA,WAAO;AAAA,EACR;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAWA,MAAc,kBAAoD;AACjE,QAAI;AACH,UAAI,KAAK,UAAU,OAAO,KAAK,OAAO,oBAAoB,YAAY;AACrE,eAAO,MAAM,KAAK,OAAO,gBAAgB;AAAA,MAC1C;AAAA,IACD,QAAQ;AAAA,IAER;AACA,WAAO,CAAC;AAAA,EACT;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAaA,MAAc,oBAAsC;AACnD,QAAI;AAGH,YAAM,SAAS,MAAM,OAAO,gBAAgB;AAC5C,aAAO;AAAA,IACR,SAAS,OAAO;AACf,YAAM,UAAU,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK;AACrE,YAAM,IAAI;AAAA,QACT,mGAAmG,OAAO;AAAA,MAC3G;AAAA,IACD;AAAA,EACD;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAWQ,eAAe,UAAwB;AAC9C,QAAI,KAAK,kBAAkB;AAC1B,UAAI;AACH,aAAK,iBAAiB,KAAK,IAAI,KAAK,KAAK,IAAI,GAAG,QAAQ,CAAC,CAAC;AAAA,MAC3D,QAAQ;AAAA,MAER;AAAA,IACD;AAAA,EACD;AACD;","names":[]}