@kreuzberg/wasm 4.3.7 → 4.4.0

package/README.md

@@ -22,7 +22,7 @@
     <img src="https://img.shields.io/maven-central/v/dev.kreuzberg/kreuzberg?label=Java&color=007ec6" alt="Java">
   </a>
   <a href="https://github.com/kreuzberg-dev/kreuzberg/releases">
-    <img src="https://img.shields.io/github/v/tag/kreuzberg-dev/kreuzberg?label=Go&color=007ec6&filter=v4.3.7" alt="Go">
+    <img src="https://img.shields.io/github/v/tag/kreuzberg-dev/kreuzberg?label=Go&color=007ec6&filter=v4.4.0" alt="Go">
   </a>
   <a href="https://www.nuget.org/packages/Kreuzberg/">
     <img src="https://img.shields.io/nuget/v/Kreuzberg?label=C%23&color=007ec6" alt="C#">
@@ -57,6 +57,8 @@
 
 Extract text, tables, images, and metadata from 75+ file formats including PDF, Office documents, and images. WebAssembly bindings for browsers, Deno, and Cloudflare Workers with portable deployment and multi-threading support.
 
+> **Full Feature Parity** — The WASM package supports all extraction capabilities at full parity with native bindings: PDF (via PDFium), Excel/spreadsheets (via Calamine), archives (ZIP, TAR, 7z, GZIP), and OCR (via built-in Tesseract-WASM). No external dependencies required.
+
 
 ## Installation
 
@@ -95,7 +97,7 @@ yarn add @kreuzberg/wasm
 ### System Requirements
 
 - Modern browser with WebAssembly support, or Deno 1.0+, or Cloudflare Workers
-- Optional: [Tesseract WASM](https://github.com/naptha/tesseract.js) for OCR functionality
+- OCR is built-in via Tesseract-WASM (enable at runtime with `enableOcr()`)
 
 
 
@@ -174,6 +176,40 @@ extractWithOcr().catch(console.error);
 See [Table Extraction Guide](https://kreuzberg.dev/features/table-extraction/) for detailed examples.
 
 
+#### Excel/Spreadsheet Extraction
+
+Extract structured data from Excel files directly in the browser or server-side runtimes:
+
+```ts
+import { extractBytes, initWasm } from "@kreuzberg/wasm";
+
+async function extractSpreadsheet() {
+	await initWasm();
+
+	const bytes = new Uint8Array(
+		await fetch("report.xlsx").then((r) => r.arrayBuffer()),
+	);
+
+	const result = await extractBytes(
+		bytes,
+		"application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
+	);
+
+	console.log("Spreadsheet content:");
+	console.log(result.content);
+
+	if (result.tables && result.tables.length > 0) {
+		result.tables.forEach((table, index) => {
+			console.log(`\nSheet ${index + 1}:`);
+			console.log(table.markdown);
+		});
+	}
+}
+
+extractSpreadsheet().catch(console.error);
+```
+
+
 
 #### Processing Multiple Files
 
@@ -318,14 +354,10 @@ extractDocuments(fileBytes, mimes)
 - **Metadata Extraction** - Retrieve document properties, creation date, author, etc.
 - **Table Extraction** - Parse tables with structure and cell content preservation
 - **Image Extraction** - Extract embedded images and render page previews
-- **OCR Support** - Integrate multiple OCR backends for scanned documents
-
+- **OCR Support** - Built-in Tesseract-WASM for scanned documents and images
+- **Full Feature Parity** - All extraction capabilities at parity with native bindings: PDF, Excel, archives, OCR, and 75+ formats
 - **Async/Await** - Non-blocking document processing with concurrent operations
-
-
 - **Plugin System** - Extensible post-processing for custom text transformation
-
-
 - **Batch Processing** - Efficiently process multiple documents in parallel
 - **Memory Efficient** - Stream large files without loading entirely into memory
 - **Language Detection** - Detect and support multiple languages in documents

package/dist/adapters/wasm-adapter.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"wasm-adapter.d.ts","sourceRoot":"","sources":["../../typescript/adapters/wasm-adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;GA0BG;AAEH,OAAO,KAAK,EAMX,gBAAgB,EAChB,gBAAgB,EAMhB,MAAM,aAAa,CAAC;AAoCrB;;;;;;;;;;;;;;;;GAgBG;AACH,wBAAsB,gBAAgB,CAAC,IAAI,EAAE,IAAI,GAAG,IAAI,GAAG,OAAO,CAAC,UAAU,CAAC,CAa7E;AAED;;;;;;;;;;;;;;;;;GAiBG;AACH,wBAAgB,UAAU,CAAC,MAAM,EAAE,gBAAgB,GAAG,IAAI,GAAG,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAoCnF;AAED;;;;;;;;;;;;;;;;GAgBG;AACH,wBAAgB,oBAAoB,CAAC,OAAO,EAAE,OAAO,GAAG,gBAAgB,CAiOvE;AAED;;;;;;;;;;;;;;;;;;;;GAoBG;AACH,wBAAgB,aAAa,CAAC,KAAK,EAAE,OAAO,EAAE,OAAO,EAAE,MAAM,GAAG,KAAK,CASpE;AAED;;;;;;;;;;GAUG;AACH,wBAAgB,uBAAuB,CAAC,KAAK,EAAE,OAAO,GAAG,KAAK,IAAI,gBAAgB,CAajF"}
+{"version":3,"file":"wasm-adapter.d.ts","sourceRoot":"","sources":["../../typescript/adapters/wasm-adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;GA0BG;AAEH,OAAO,KAAK,EAMX,gBAAgB,EAChB,gBAAgB,EAOhB,MAAM,aAAa,CAAC;AAoCrB;;;;;;;;;;;;;;;;GAgBG;AACH,wBAAsB,gBAAgB,CAAC,IAAI,EAAE,IAAI,GAAG,IAAI,GAAG,OAAO,CAAC,UAAU,CAAC,CAa7E;AAED;;;;;;;;;;;;;;;;;GAiBG;AACH,wBAAgB,UAAU,CAAC,MAAM,EAAE,gBAAgB,GAAG,IAAI,GAAG,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAsCnF;AAED;;;;;;;;;;;;;;;;GAgBG;AACH,wBAAgB,oBAAoB,CAAC,OAAO,EAAE,OAAO,GAAG,gBAAgB,CAwOvE;AAED;;;;;;;;;;;;;;;;;;;;GAoBG;AACH,wBAAgB,aAAa,CAAC,KAAK,EAAE,OAAO,EAAE,OAAO,EAAE,MAAM,GAAG,KAAK,CASpE;AAED;;;;;;;;;;GAUG;AACH,wBAAgB,uBAAuB,CAAC,KAAK,EAAE,OAAO,GAAG,KAAK,IAAI,gBAAgB,CAajF"}

package/dist/adapters/wasm-adapter.js

@@ -26,7 +26,7 @@ function configToJS(config) {
   if (!config) {
     return {};
   }
-  const normalized = {};
+  const toSnakeCase = (str) => str.replace(/[A-Z]/g, (letter) => `_${letter.toLowerCase()}`);
   const normalizeValue = (value) => {
     if (value === null || value === void 0) {
       return null;
@@ -40,17 +40,18 @@ function configToJS(config) {
       for (const [key, val] of Object.entries(obj)) {
         const normalizedVal = normalizeValue(val);
         if (normalizedVal !== null && normalizedVal !== void 0) {
-          normalized2[key] = normalizedVal;
+          normalized2[toSnakeCase(key)] = normalizedVal;
         }
       }
       return Object.keys(normalized2).length > 0 ? normalized2 : null;
     }
     return value;
   };
+  const normalized = {};
   for (const [key, value] of Object.entries(config)) {
     const normalizedValue = normalizeValue(value);
     if (normalizedValue !== null && normalizedValue !== void 0) {
-      normalized[key] = normalizedValue;
+      normalized[toSnakeCase(key)] = normalizedValue;
     }
   }
   return normalized;
@@ -153,8 +154,13 @@ function jsToExtractionResult(jsValue) {
       throw new Error("Invalid image structure");
     }
     const img = image;
-    if (!(img.data instanceof Uint8Array)) {
-      throw new Error("Invalid image: data must be Uint8Array");
+    let imageData;
+    if (img.data instanceof Uint8Array) {
+      imageData = img.data;
+    } else if (Array.isArray(img.data)) {
+      imageData = new Uint8Array(img.data);
+    } else {
+      throw new Error("Invalid image: data must be Uint8Array or array");
     }
     if (typeof img.format !== "string") {
       throw new Error("Invalid image: missing format");
@@ -189,7 +195,7 @@ function jsToExtractionResult(jsValue) {
       throw new Error("Invalid image: description must be a string or null");
     }
     return {
-      data: img.data,
+      data: imageData,
       format: img.format,
       imageIndex,
       pageNumber: pageNumber ?? null,
@@ -217,6 +223,7 @@ function jsToExtractionResult(jsValue) {
   const ocrElements = result.ocrElements ?? result.ocr_elements ?? null;
   const document = result.document ?? null;
   const pages = result.pages ?? null;
+  const annotations = result.annotations ?? null;
   return {
     content: result.content,
     mimeType,
@@ -231,7 +238,8 @@ function jsToExtractionResult(jsValue) {
     processingWarnings,
     elements,
     ocrElements,
-    document
+    document,
+    annotations
   };
 }
 function wrapWasmError(error, context) {

package/dist/adapters/wasm-adapter.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../../typescript/adapters/wasm-adapter.ts"],"sourcesContent":["/**\n * WASM Type Adapter\n *\n * This module provides type adapters for converting between JavaScript/TypeScript\n * types and WASM-compatible types, handling File/Blob conversions, config normalization,\n * and result parsing.\n *\n * @example File Conversion\n * ```typescript\n * import { fileToUint8Array } from '@kreuzberg/wasm/adapters/wasm-adapter';\n *\n * const file = event.target.files[0];\n * const bytes = await fileToUint8Array(file);\n * const result = await extractBytes(bytes, file.type);\n * ```\n *\n * @example Config Normalization\n * ```typescript\n * import { configToJS } from '@kreuzberg/wasm/adapters/wasm-adapter';\n *\n * const config = {\n *   ocr: { backend: 'tesseract', language: 'eng' },\n *   chunking: { maxChars: 1000 }\n * };\n * const normalized = configToJS(config);\n * ```\n */\n\nimport type {\n\tChunk,\n\tDocumentStructure,\n\tElement,\n\tExtractedImage,\n\tExtractedKeyword,\n\tExtractionConfig,\n\tExtractionResult,\n\tMetadata,\n\tOcrElement,\n\tPageContent,\n\tProcessingWarning,\n\tTable,\n} from \"../types.js\";\n\n/**\n * Maximum file size for processing (512 MB)\n *\n * @internal\n */\nconst MAX_FILE_SIZE = 512 * 1024 * 1024;\n\n/**\n * Type predicate to validate numeric value or null\n *\n * @internal\n */\nfunction isNumberOrNull(value: unknown): value is number | null {\n\treturn typeof value === \"number\" || value === null || value === undefined;\n}\n\n/**\n * Type predicate to validate string value or null\n *\n * @internal\n */\nfunction isStringOrNull(value: unknown): value is string | null {\n\treturn typeof value === \"string\" || value === null || value === undefined;\n}\n\n/**\n * Type predicate to validate boolean value\n *\n * @internal\n */\nfunction isBoolean(value: unknown): value is boolean {\n\treturn typeof value === \"boolean\" || value === undefined;\n}\n\n/**\n * Convert a File or Blob to Uint8Array\n *\n * Handles both browser File API and server-side Blob-like objects,\n * providing a unified interface for reading binary data.\n *\n * @param file - The File or Blob to convert\n * @returns Promise resolving to the byte array\n * @throws {Error} If the file cannot be read or exceeds size limit\n *\n * @example\n * ```typescript\n * const file = document.getElementById('input').files[0];\n * const bytes = await fileToUint8Array(file);\n * const result = await extractBytes(bytes, 'application/pdf');\n * ```\n */\nexport async function fileToUint8Array(file: File | Blob): Promise<Uint8Array> {\n\ttry {\n\t\tif (file.size > MAX_FILE_SIZE) {\n\t\t\tthrow new Error(\n\t\t\t\t`File size (${file.size} bytes) exceeds maximum (${MAX_FILE_SIZE} bytes). Maximum file size is 512 MB.`,\n\t\t\t);\n\t\t}\n\n\t\tconst arrayBuffer = await file.arrayBuffer();\n\t\treturn new Uint8Array(arrayBuffer);\n\t} catch (error) {\n\t\tthrow new Error(`Failed to read file: ${error instanceof Error ? error.message : String(error)}`);\n\t}\n}\n\n/**\n * Normalize ExtractionConfig for WASM processing\n *\n * Converts TypeScript configuration objects to a WASM-compatible format,\n * handling null values, undefined properties, and nested structures.\n *\n * @param config - The extraction configuration or null\n * @returns Normalized configuration object suitable for WASM\n *\n * @example\n * ```typescript\n * const config: ExtractionConfig = {\n *   ocr: { backend: 'tesseract' },\n *   chunking: { maxChars: 1000 }\n * };\n * const wasmConfig = configToJS(config);\n * ```\n */\nexport function configToJS(config: ExtractionConfig | null): Record<string, unknown> {\n\tif (!config) {\n\t\treturn {};\n\t}\n\n\tconst normalized: Record<string, unknown> = {};\n\n\tconst normalizeValue = (value: unknown): unknown => {\n\t\tif (value === null || value === undefined) {\n\t\t\treturn null;\n\t\t}\n\t\tif (typeof value === \"object\") {\n\t\t\tif (Array.isArray(value)) {\n\t\t\t\treturn value.map(normalizeValue);\n\t\t\t}\n\t\t\tconst obj = value as Record<string, unknown>;\n\t\t\tconst normalized: Record<string, unknown> = {};\n\t\t\tfor (const [key, val] of Object.entries(obj)) {\n\t\t\t\tconst normalizedVal = normalizeValue(val);\n\t\t\t\tif (normalizedVal !== null && normalizedVal !== undefined) {\n\t\t\t\t\tnormalized[key] = normalizedVal;\n\t\t\t\t}\n\t\t\t}\n\t\t\treturn Object.keys(normalized).length > 0 ? normalized : null;\n\t\t}\n\t\treturn value;\n\t};\n\n\tfor (const [key, value] of Object.entries(config)) {\n\t\tconst normalizedValue = normalizeValue(value);\n\t\tif (normalizedValue !== null && normalizedValue !== undefined) {\n\t\t\tnormalized[key] = normalizedValue;\n\t\t}\n\t}\n\n\treturn normalized;\n}\n\n/**\n * Parse WASM extraction result and convert to TypeScript type\n *\n * Handles conversion of WASM-returned objects to proper ExtractionResult types,\n * including proper array conversions and type assertions for tables, chunks, and images.\n *\n * @param jsValue - The raw WASM result value\n * @returns Properly typed ExtractionResult\n * @throws {Error} If the result structure is invalid\n *\n * @example\n * ```typescript\n * const wasmResult = await wasmExtract(bytes, mimeType, config);\n * const result = jsToExtractionResult(wasmResult);\n * console.log(result.content);\n * ```\n */\nexport function jsToExtractionResult(jsValue: unknown): ExtractionResult {\n\tif (!jsValue || typeof jsValue !== \"object\") {\n\t\tthrow new Error(\"Invalid extraction result: value is not an object\");\n\t}\n\n\tconst result = jsValue as Record<string, unknown>;\n\tconst mimeType =\n\t\ttypeof result.mimeType === \"string\"\n\t\t\t? result.mimeType\n\t\t\t: typeof result.mime_type === \"string\"\n\t\t\t\t? result.mime_type\n\t\t\t\t: null;\n\n\tif (typeof result.content !== \"string\") {\n\t\tthrow new Error(\"Invalid extraction result: missing or invalid content\");\n\t}\n\tif (typeof mimeType !== \"string\") {\n\t\tthrow new Error(\"Invalid extraction result: missing or invalid mimeType\");\n\t}\n\tif (!result.metadata || typeof result.metadata !== \"object\") {\n\t\tthrow new Error(\"Invalid extraction result: missing or invalid metadata\");\n\t}\n\n\tconst tables: Table[] = [];\n\tif (Array.isArray(result.tables)) {\n\t\tfor (const table of result.tables) {\n\t\t\tif (table && typeof table === \"object\") {\n\t\t\t\tconst t = table as Record<string, unknown>;\n\t\t\t\tconst pageNumber =\n\t\t\t\t\ttypeof t.pageNumber === \"number\" ? t.pageNumber : typeof t.page_number === \"number\" ? t.page_number : null;\n\t\t\t\tif (\n\t\t\t\t\tArray.isArray(t.cells) &&\n\t\t\t\t\tt.cells.every((row) => Array.isArray(row) && row.every((cell) => typeof cell === \"string\")) &&\n\t\t\t\t\ttypeof t.markdown === \"string\" &&\n\t\t\t\t\tpageNumber !== null\n\t\t\t\t) {\n\t\t\t\t\ttables.push({\n\t\t\t\t\t\tcells: t.cells as string[][],\n\t\t\t\t\t\tmarkdown: t.markdown,\n\t\t\t\t\t\tpageNumber,\n\t\t\t\t\t});\n\t\t\t\t}\n\t\t\t}\n\t\t}\n\t}\n\n\tconst chunks: Chunk[] | null = Array.isArray(result.chunks)\n\t\t? result.chunks.map((chunk) => {\n\t\t\t\tif (!chunk || typeof chunk !== \"object\") {\n\t\t\t\t\tthrow new Error(\"Invalid chunk structure\");\n\t\t\t\t}\n\t\t\t\tconst c = chunk as Record<string, unknown>;\n\t\t\t\tif (typeof c.content !== \"string\") {\n\t\t\t\t\tthrow new Error(\"Invalid chunk: missing content\");\n\t\t\t\t}\n\t\t\t\tif (!c.metadata || typeof c.metadata !== \"object\") {\n\t\t\t\t\tthrow new Error(\"Invalid chunk: missing metadata\");\n\t\t\t\t}\n\t\t\t\tconst metadata = c.metadata as Record<string, unknown>;\n\n\t\t\t\tlet embedding: number[] | null = null;\n\t\t\t\tif (Array.isArray(c.embedding)) {\n\t\t\t\t\tif (!c.embedding.every((item) => typeof item === \"number\")) {\n\t\t\t\t\t\tthrow new Error(\"Invalid chunk: embedding must contain only numbers\");\n\t\t\t\t\t}\n\t\t\t\t\tembedding = c.embedding;\n\t\t\t\t}\n\n\t\t\t\t// Coerce numeric values - handle BigInt, strings, and numbers\n\t\t\t\tconst coerceToNumber = (value: unknown, fieldName: string): number => {\n\t\t\t\t\tif (typeof value === \"number\") {\n\t\t\t\t\t\treturn value;\n\t\t\t\t\t}\n\t\t\t\t\tif (typeof value === \"bigint\") {\n\t\t\t\t\t\treturn Number(value);\n\t\t\t\t\t}\n\t\t\t\t\tif (typeof value === \"string\") {\n\t\t\t\t\t\tconst parsed = parseInt(value, 10);\n\t\t\t\t\t\tif (Number.isNaN(parsed)) {\n\t\t\t\t\t\t\tthrow new Error(`Invalid chunk metadata: ${fieldName} must be a valid number, got \"${value}\"`);\n\t\t\t\t\t\t}\n\t\t\t\t\t\treturn parsed;\n\t\t\t\t\t}\n\t\t\t\t\tthrow new Error(`Invalid chunk metadata: ${fieldName} must be a number, got ${typeof value}`);\n\t\t\t\t};\n\n\t\t\t\t// The Rust code uses snake_case field names (byte_start, byte_end, etc)\n\t\t\t\t// but TypeScript expects camelCase (charStart, charEnd, etc)\n\t\t\t\t// For now, treat byte offsets as character offsets since the content is UTF-8\n\t\t\t\tconst charStart = coerceToNumber(\n\t\t\t\t\tmetadata.charStart ?? metadata.char_start ?? metadata.byteStart ?? metadata.byte_start,\n\t\t\t\t\t\"charStart\",\n\t\t\t\t);\n\t\t\t\tconst charEnd = coerceToNumber(\n\t\t\t\t\tmetadata.charEnd ?? metadata.char_end ?? metadata.byteEnd ?? metadata.byte_end,\n\t\t\t\t\t\"charEnd\",\n\t\t\t\t);\n\t\t\t\tconst chunkIndex = coerceToNumber(metadata.chunkIndex ?? metadata.chunk_index, \"chunkIndex\");\n\t\t\t\tconst totalChunks = coerceToNumber(metadata.totalChunks ?? metadata.total_chunks, \"totalChunks\");\n\n\t\t\t\tlet tokenCount: number | null = null;\n\t\t\t\tconst tokenCountValue = metadata.tokenCount ?? metadata.token_count;\n\t\t\t\tif (tokenCountValue !== null && tokenCountValue !== undefined) {\n\t\t\t\t\ttokenCount = coerceToNumber(tokenCountValue, \"tokenCount\");\n\t\t\t\t}\n\n\t\t\t\treturn {\n\t\t\t\t\tcontent: c.content,\n\t\t\t\t\tembedding,\n\t\t\t\t\tmetadata: {\n\t\t\t\t\t\tcharStart,\n\t\t\t\t\t\tcharEnd,\n\t\t\t\t\t\ttokenCount,\n\t\t\t\t\t\tchunkIndex,\n\t\t\t\t\t\ttotalChunks,\n\t\t\t\t\t},\n\t\t\t\t};\n\t\t\t})\n\t\t: null;\n\n\tconst images: ExtractedImage[] | null = Array.isArray(result.images)\n\t\t? result.images.map((image) => {\n\t\t\t\tif (!image || typeof image !== \"object\") {\n\t\t\t\t\tthrow new Error(\"Invalid image structure\");\n\t\t\t\t}\n\t\t\t\tconst img = image as Record<string, unknown>;\n\t\t\t\tif (!(img.data instanceof Uint8Array)) {\n\t\t\t\t\tthrow new Error(\"Invalid image: data must be Uint8Array\");\n\t\t\t\t}\n\t\t\t\tif (typeof img.format !== \"string\") {\n\t\t\t\t\tthrow new Error(\"Invalid image: missing format\");\n\t\t\t\t}\n\n\t\t\t\t// Support both camelCase and snake_case field names (Rust serde uses snake_case)\n\t\t\t\tconst imageIndex = img.imageIndex ?? img.image_index;\n\t\t\t\tconst pageNumber = img.pageNumber ?? img.page_number;\n\t\t\t\tconst bitsPerComponent = img.bitsPerComponent ?? img.bits_per_component;\n\t\t\t\tconst isMask = img.isMask ?? img.is_mask;\n\t\t\t\tconst ocrResult = img.ocrResult ?? img.ocr_result;\n\n\t\t\t\tif (typeof imageIndex !== \"number\") {\n\t\t\t\t\tthrow new Error(\"Invalid image: imageIndex must be a number\");\n\t\t\t\t}\n\t\t\t\tif (!isNumberOrNull(pageNumber)) {\n\t\t\t\t\tthrow new Error(\"Invalid image: pageNumber must be a number or null\");\n\t\t\t\t}\n\t\t\t\tif (!isNumberOrNull(img.width)) {\n\t\t\t\t\tthrow new Error(\"Invalid image: width must be a number or null\");\n\t\t\t\t}\n\t\t\t\tif (!isNumberOrNull(img.height)) {\n\t\t\t\t\tthrow new Error(\"Invalid image: height must be a number or null\");\n\t\t\t\t}\n\t\t\t\tif (!isNumberOrNull(bitsPerComponent)) {\n\t\t\t\t\tthrow new Error(\"Invalid image: bitsPerComponent must be a number or null\");\n\t\t\t\t}\n\n\t\t\t\tif (!isBoolean(isMask)) {\n\t\t\t\t\tthrow new Error(\"Invalid image: isMask must be a boolean\");\n\t\t\t\t}\n\n\t\t\t\tif (!isStringOrNull(img.colorspace)) {\n\t\t\t\t\tthrow new Error(\"Invalid image: colorspace must be a string or null\");\n\t\t\t\t}\n\t\t\t\tif (!isStringOrNull(img.description)) {\n\t\t\t\t\tthrow new Error(\"Invalid image: description must be a string or null\");\n\t\t\t\t}\n\n\t\t\t\treturn {\n\t\t\t\t\tdata: img.data,\n\t\t\t\t\tformat: img.format,\n\t\t\t\t\timageIndex: imageIndex,\n\t\t\t\t\tpageNumber: pageNumber ?? null,\n\t\t\t\t\twidth: (img.width as number) ?? null,\n\t\t\t\t\theight: (img.height as number) ?? null,\n\t\t\t\t\tcolorspace: (img.colorspace as string) ?? null,\n\t\t\t\t\tbitsPerComponent: bitsPerComponent ?? null,\n\t\t\t\t\tisMask: isMask ?? false,\n\t\t\t\t\tdescription: (img.description as string) ?? null,\n\t\t\t\t\tocrResult: ocrResult ? jsToExtractionResult(ocrResult) : null,\n\t\t\t\t};\n\t\t\t})\n\t\t: null;\n\n\tlet detectedLanguages: string[] | null = null;\n\tconst detectedLanguagesRaw = Array.isArray(result.detectedLanguages)\n\t\t? result.detectedLanguages\n\t\t: result.detected_languages;\n\tif (Array.isArray(detectedLanguagesRaw)) {\n\t\tif (!detectedLanguagesRaw.every((lang) => typeof lang === \"string\")) {\n\t\t\tthrow new Error(\"Invalid result: detectedLanguages must contain only strings\");\n\t\t}\n\t\tdetectedLanguages = detectedLanguagesRaw;\n\t}\n\n\tconst extractedKeywords = (result.extractedKeywords ?? result.extracted_keywords ?? null) as\n\t\t| ExtractedKeyword[]\n\t\t| null;\n\tconst qualityScore =\n\t\ttypeof (result.qualityScore ?? result.quality_score) === \"number\"\n\t\t\t? ((result.qualityScore ?? result.quality_score) as number)\n\t\t\t: null;\n\tconst processingWarnings = (result.processingWarnings ?? result.processing_warnings ?? null) as\n\t\t| ProcessingWarning[]\n\t\t| null;\n\tconst elements = (result.elements ?? null) as Element[] | null;\n\tconst ocrElements = (result.ocrElements ?? result.ocr_elements ?? null) as OcrElement[] | null;\n\tconst document = (result.document ?? null) as DocumentStructure | null;\n\tconst pages = (result.pages ?? null) as PageContent[] | null;\n\n\treturn {\n\t\tcontent: result.content,\n\t\tmimeType,\n\t\tmetadata: (result.metadata ?? {}) as Metadata,\n\t\ttables,\n\t\tdetectedLanguages,\n\t\tchunks,\n\t\timages,\n\t\tpages,\n\t\textractedKeywords,\n\t\tqualityScore,\n\t\tprocessingWarnings,\n\t\telements,\n\t\tocrElements,\n\t\tdocument,\n\t};\n}\n\n/**\n * Wrap and format WASM errors with context\n *\n * Converts WASM error messages to JavaScript Error objects with proper context\n * and stack trace information when available.\n *\n * @param error - The error from WASM\n * @param context - Additional context about what operation failed\n * @returns A formatted Error object\n *\n * @internal\n *\n * @example\n * ```typescript\n * try {\n *   await wasmExtract(bytes, mimeType);\n * } catch (error) {\n *   throw wrapWasmError(error, 'extracting document');\n * }\n * ```\n */\nexport function wrapWasmError(error: unknown, context: string): Error {\n\tif (error instanceof Error) {\n\t\treturn new Error(`Error ${context}: ${error.message}`, {\n\t\t\tcause: error,\n\t\t});\n\t}\n\n\tconst message = String(error);\n\treturn new Error(`Error ${context}: ${message}`);\n}\n\n/**\n * Validate that a WASM-returned value conforms to ExtractionResult structure\n *\n * Performs structural validation without full type checking,\n * useful for runtime validation of WASM output.\n *\n * @param value - The value to validate\n * @returns True if value appears to be a valid ExtractionResult\n *\n * @internal\n */\nexport function isValidExtractionResult(value: unknown): value is ExtractionResult {\n\tif (!value || typeof value !== \"object\") {\n\t\treturn false;\n\t}\n\n\tconst obj = value as Record<string, unknown>;\n\treturn (\n\t\ttypeof obj.content === \"string\" &&\n\t\t(typeof obj.mimeType === \"string\" || typeof obj.mime_type === \"string\") &&\n\t\tobj.metadata !== null &&\n\t\ttypeof obj.metadata === \"object\" &&\n\t\tArray.isArray(obj.tables)\n\t);\n}\n"],"mappings":";AAgDA,IAAM,gBAAgB,MAAM,OAAO;AAOnC,SAAS,eAAe,OAAwC;AAC/D,SAAO,OAAO,UAAU,YAAY,UAAU,QAAQ,UAAU;AACjE;AAOA,SAAS,eAAe,OAAwC;AAC/D,SAAO,OAAO,UAAU,YAAY,UAAU,QAAQ,UAAU;AACjE;AAOA,SAAS,UAAU,OAAkC;AACpD,SAAO,OAAO,UAAU,aAAa,UAAU;AAChD;AAmBA,eAAsB,iBAAiB,MAAwC;AAC9E,MAAI;AACH,QAAI,KAAK,OAAO,eAAe;AAC9B,YAAM,IAAI;AAAA,QACT,cAAc,KAAK,IAAI,4BAA4B,aAAa;AAAA,MACjE;AAAA,IACD;AAEA,UAAM,cAAc,MAAM,KAAK,YAAY;AAC3C,WAAO,IAAI,WAAW,WAAW;AAAA,EAClC,SAAS,OAAO;AACf,UAAM,IAAI,MAAM,wBAAwB,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK,CAAC,EAAE;AAAA,EACjG;AACD;AAoBO,SAAS,WAAW,QAA0D;AACpF,MAAI,CAAC,QAAQ;AACZ,WAAO,CAAC;AAAA,EACT;AAEA,QAAM,aAAsC,CAAC;AAE7C,QAAM,iBAAiB,CAAC,UAA4B;AACnD,QAAI,UAAU,QAAQ,UAAU,QAAW;AAC1C,aAAO;AAAA,IACR;AACA,QAAI,OAAO,UAAU,UAAU;AAC9B,UAAI,MAAM,QAAQ,KAAK,GAAG;AACzB,eAAO,MAAM,IAAI,cAAc;AAAA,MAChC;AACA,YAAM,MAAM;AACZ,YAAMA,cAAsC,CAAC;AAC7C,iBAAW,CAAC,KAAK,GAAG,KAAK,OAAO,QAAQ,GAAG,GAAG;AAC7C,cAAM,gBAAgB,eAAe,GAAG;AACxC,YAAI,kBAAkB,QAAQ,kBAAkB,QAAW;AAC1D,UAAAA,YAAW,GAAG,IAAI;AAAA,QACnB;AAAA,MACD;AACA,aAAO,OAAO,KAAKA,WAAU,EAAE,SAAS,IAAIA,cAAa;AAAA,IAC1D;AACA,WAAO;AAAA,EACR;AAEA,aAAW,CAAC,KAAK,KAAK,KAAK,OAAO,QAAQ,MAAM,GAAG;AAClD,UAAM,kBAAkB,eAAe,KAAK;AAC5C,QAAI,oBAAoB,QAAQ,oBAAoB,QAAW;AAC9D,iBAAW,GAAG,IAAI;AAAA,IACnB;AAAA,EACD;AAEA,SAAO;AACR;AAmBO,SAAS,qBAAqB,SAAoC;AACxE,MAAI,CAAC,WAAW,OAAO,YAAY,UAAU;AAC5C,UAAM,IAAI,MAAM,mDAAmD;AAAA,EACpE;AAEA,QAAM,SAAS;AACf,QAAM,WACL,OAAO,OAAO,aAAa,WACxB,OAAO,WACP,OAAO,OAAO,cAAc,WAC3B,OAAO,YACP;AAEL,MAAI,OAAO,OAAO,YAAY,UAAU;AACvC,UAAM,IAAI,MAAM,uDAAuD;AAAA,EACxE;AACA,MAAI,OAAO,aAAa,UAAU;AACjC,UAAM,IAAI,MAAM,wDAAwD;AAAA,EACzE;AACA,MAAI,CAAC,OAAO,YAAY,OAAO,OAAO,aAAa,UAAU;AAC5D,UAAM,IAAI,MAAM,wDAAwD;AAAA,EACzE;AAEA,QAAM,SAAkB,CAAC;AACzB,MAAI,MAAM,QAAQ,OAAO,MAAM,GAAG;AACjC,eAAW,SAAS,OAAO,QAAQ;AAClC,UAAI,SAAS,OAAO,UAAU,UAAU;AACvC,cAAM,IAAI;AACV,cAAM,aACL,OAAO,EAAE,eAAe,WAAW,EAAE,aAAa,OAAO,EAAE,gBAAgB,WAAW,EAAE,cAAc;AACvG,YACC,MAAM,QAAQ,EAAE,KAAK,KACrB,EAAE,MAAM,MAAM,CAAC,QAAQ,MAAM,QAAQ,GAAG,KAAK,IAAI,MAAM,CAAC,SAAS,OAAO,SAAS,QAAQ,CAAC,KAC1F,OAAO,EAAE,aAAa,YACtB,eAAe,MACd;AACD,iBAAO,KAAK;AAAA,YACX,OAAO,EAAE;AAAA,YACT,UAAU,EAAE;AAAA,YACZ;AAAA,UACD,CAAC;AAAA,QACF;AAAA,MACD;AAAA,IACD;AAAA,EACD;AAEA,QAAM,SAAyB,MAAM,QAAQ,OAAO,MAAM,IACvD,OAAO,OAAO,IAAI,CAAC,UAAU;AAC7B,QAAI,CAAC,SAAS,OAAO,UAAU,UAAU;AACxC,YAAM,IAAI,MAAM,yBAAyB;AAAA,IAC1C;AACA,UAAM,IAAI;AACV,QAAI,OAAO,EAAE,YAAY,UAAU;AAClC,YAAM,IAAI,MAAM,gCAAgC;AAAA,IACjD;AACA,QAAI,CAAC,EAAE,YAAY,OAAO,EAAE,aAAa,UAAU;AAClD,YAAM,IAAI,MAAM,iCAAiC;AAAA,IAClD;AACA,UAAM,WAAW,EAAE;AAEnB,QAAI,YAA6B;AACjC,QAAI,MAAM,QAAQ,EAAE,SAAS,GAAG;AAC/B,UAAI,CAAC,EAAE,UAAU,MAAM,CAAC,SAAS,OAAO,SAAS,QAAQ,GAAG;AAC3D,cAAM,IAAI,MAAM,oDAAoD;AAAA,MACrE;AACA,kBAAY,EAAE;AAAA,IACf;AAGA,UAAM,iBAAiB,CAAC,OAAgB,cAA8B;AACrE,UAAI,OAAO,UAAU,UAAU;AAC9B,eAAO;AAAA,MACR;AACA,UAAI,OAAO,UAAU,UAAU;AAC9B,eAAO,OAAO,KAAK;AAAA,MACpB;AACA,UAAI,OAAO,UAAU,UAAU;AAC9B,cAAM,SAAS,SAAS,OAAO,EAAE;AACjC,YAAI,OAAO,MAAM,MAAM,GAAG;AACzB,gBAAM,IAAI,MAAM,2BAA2B,SAAS,iCAAiC,KAAK,GAAG;AAAA,QAC9F;AACA,eAAO;AAAA,MACR;AACA,YAAM,IAAI,MAAM,2BAA2B,SAAS,0BAA0B,OAAO,KAAK,EAAE;AAAA,IAC7F;AAKA,UAAM,YAAY;AAAA,MACjB,SAAS,aAAa,SAAS,cAAc,SAAS,aAAa,SAAS;AAAA,MAC5E;AAAA,IACD;AACA,UAAM,UAAU;AAAA,MACf,SAAS,WAAW,SAAS,YAAY,SAAS,WAAW,SAAS;AAAA,MACtE;AAAA,IACD;AACA,UAAM,aAAa,eAAe,SAAS,cAAc,SAAS,aAAa,YAAY;AAC3F,UAAM,cAAc,eAAe,SAAS,eAAe,SAAS,cAAc,aAAa;AAE/F,QAAI,aAA4B;AAChC,UAAM,kBAAkB,SAAS,cAAc,SAAS;AACxD,QAAI,oBAAoB,QAAQ,oBAAoB,QAAW;AAC9D,mBAAa,eAAe,iBAAiB,YAAY;AAAA,IAC1D;AAEA,WAAO;AAAA,MACN,SAAS,EAAE;AAAA,MACX;AAAA,MACA,UAAU;AAAA,QACT;AAAA,QACA;AAAA,QACA;AAAA,QACA;AAAA,QACA;AAAA,MACD;AAAA,IACD;AAAA,EACD,CAAC,IACA;AAEH,QAAM,SAAkC,MAAM,QAAQ,OAAO,MAAM,IAChE,OAAO,OAAO,IAAI,CAAC,UAAU;AAC7B,QAAI,CAAC,SAAS,OAAO,UAAU,UAAU;AACxC,YAAM,IAAI,MAAM,yBAAyB;AAAA,IAC1C;AACA,UAAM,MAAM;AACZ,QAAI,EAAE,IAAI,gBAAgB,aAAa;AACtC,YAAM,IAAI,MAAM,wCAAwC;AAAA,IACzD;AACA,QAAI,OAAO,IAAI,WAAW,UAAU;AACnC,YAAM,IAAI,MAAM,+BAA+B;AAAA,IAChD;AAGA,UAAM,aAAa,IAAI,cAAc,IAAI;AACzC,UAAM,aAAa,IAAI,cAAc,IAAI;AACzC,UAAM,mBAAmB,IAAI,oBAAoB,IAAI;AACrD,UAAM,SAAS,IAAI,UAAU,IAAI;AACjC,UAAM,YAAY,IAAI,aAAa,IAAI;AAEvC,QAAI,OAAO,eAAe,UAAU;AACnC,YAAM,IAAI,MAAM,4CAA4C;AAAA,IAC7D;AACA,QAAI,CAAC,eAAe,UAAU,GAAG;AAChC,YAAM,IAAI,MAAM,oDAAoD;AAAA,IACrE;AACA,QAAI,CAAC,eAAe,IAAI,KAAK,GAAG;AAC/B,YAAM,IAAI,MAAM,+CAA+C;AAAA,IAChE;AACA,QAAI,CAAC,eAAe,IAAI,MAAM,GAAG;AAChC,YAAM,IAAI,MAAM,gDAAgD;AAAA,IACjE;AACA,QAAI,CAAC,eAAe,gBAAgB,GAAG;AACtC,YAAM,IAAI,MAAM,0DAA0D;AAAA,IAC3E;AAEA,QAAI,CAAC,UAAU,MAAM,GAAG;AACvB,YAAM,IAAI,MAAM,yCAAyC;AAAA,IAC1D;AAEA,QAAI,CAAC,eAAe,IAAI,UAAU,GAAG;AACpC,YAAM,IAAI,MAAM,oDAAoD;AAAA,IACrE;AACA,QAAI,CAAC,eAAe,IAAI,WAAW,GAAG;AACrC,YAAM,IAAI,MAAM,qDAAqD;AAAA,IACtE;AAEA,WAAO;AAAA,MACN,MAAM,IAAI;AAAA,MACV,QAAQ,IAAI;AAAA,MACZ;AAAA,MACA,YAAY,cAAc;AAAA,MAC1B,OAAQ,IAAI,SAAoB;AAAA,MAChC,QAAS,IAAI,UAAqB;AAAA,MAClC,YAAa,IAAI,cAAyB;AAAA,MAC1C,kBAAkB,oBAAoB;AAAA,MACtC,QAAQ,UAAU;AAAA,MAClB,aAAc,IAAI,eAA0B;AAAA,MAC5C,WAAW,YAAY,qBAAqB,SAAS,IAAI;AAAA,IAC1D;AAAA,EACD,CAAC,IACA;AAEH,MAAI,oBAAqC;AACzC,QAAM,uBAAuB,MAAM,QAAQ,OAAO,iBAAiB,IAChE,OAAO,oBACP,OAAO;AACV,MAAI,MAAM,QAAQ,oBAAoB,GAAG;AACxC,QAAI,CAAC,qBAAqB,MAAM,CAAC,SAAS,OAAO,SAAS,QAAQ,GAAG;AACpE,YAAM,IAAI,MAAM,6DAA6D;AAAA,IAC9E;AACA,wBAAoB;AAAA,EACrB;AAEA,QAAM,oBAAqB,OAAO,qBAAqB,OAAO,sBAAsB;AAGpF,QAAM,eACL,QAAQ,OAAO,gBAAgB,OAAO,mBAAmB,WACpD,OAAO,gBAAgB,OAAO,gBAChC;AACJ,QAAM,qBAAsB,OAAO,sBAAsB,OAAO,uBAAuB;AAGvF,QAAM,WAAY,OAAO,YAAY;AACrC,QAAM,cAAe,OAAO,eAAe,OAAO,gBAAgB;AAClE,QAAM,WAAY,OAAO,YAAY;AACrC,QAAM,QAAS,OAAO,SAAS;AAE/B,SAAO;AAAA,IACN,SAAS,OAAO;AAAA,IAChB;AAAA,IACA,UAAW,OAAO,YAAY,CAAC;AAAA,IAC/B;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,EACD;AACD;AAuBO,SAAS,cAAc,OAAgB,SAAwB;AACrE,MAAI,iBAAiB,OAAO;AAC3B,WAAO,IAAI,MAAM,SAAS,OAAO,KAAK,MAAM,OAAO,IAAI;AAAA,MACtD,OAAO;AAAA,IACR,CAAC;AAAA,EACF;AAEA,QAAM,UAAU,OAAO,KAAK;AAC5B,SAAO,IAAI,MAAM,SAAS,OAAO,KAAK,OAAO,EAAE;AAChD;AAaO,SAAS,wBAAwB,OAA2C;AAClF,MAAI,CAAC,SAAS,OAAO,UAAU,UAAU;AACxC,WAAO;AAAA,EACR;AAEA,QAAM,MAAM;AACZ,SACC,OAAO,IAAI,YAAY,aACtB,OAAO,IAAI,aAAa,YAAY,OAAO,IAAI,cAAc,aAC9D,IAAI,aAAa,QACjB,OAAO,IAAI,aAAa,YACxB,MAAM,QAAQ,IAAI,MAAM;AAE1B;","names":["normalized"]}
+{"version":3,"sources":["../../typescript/adapters/wasm-adapter.ts"],"sourcesContent":["/**\n * WASM Type Adapter\n *\n * This module provides type adapters for converting between JavaScript/TypeScript\n * types and WASM-compatible types, handling File/Blob conversions, config normalization,\n * and result parsing.\n *\n * @example File Conversion\n * ```typescript\n * import { fileToUint8Array } from '@kreuzberg/wasm/adapters/wasm-adapter';\n *\n * const file = event.target.files[0];\n * const bytes = await fileToUint8Array(file);\n * const result = await extractBytes(bytes, file.type);\n * ```\n *\n * @example Config Normalization\n * ```typescript\n * import { configToJS } from '@kreuzberg/wasm/adapters/wasm-adapter';\n *\n * const config = {\n *   ocr: { backend: 'tesseract', language: 'eng' },\n *   chunking: { maxChars: 1000 }\n * };\n * const normalized = configToJS(config);\n * ```\n */\n\nimport type {\n\tChunk,\n\tDocumentStructure,\n\tElement,\n\tExtractedImage,\n\tExtractedKeyword,\n\tExtractionConfig,\n\tExtractionResult,\n\tMetadata,\n\tOcrElement,\n\tPageContent,\n\tPdfAnnotation,\n\tProcessingWarning,\n\tTable,\n} from \"../types.js\";\n\n/**\n * Maximum file size for processing (512 MB)\n *\n * @internal\n */\nconst MAX_FILE_SIZE = 512 * 1024 * 1024;\n\n/**\n * Type predicate to validate numeric value or null\n *\n * @internal\n */\nfunction isNumberOrNull(value: unknown): value is number | null {\n\treturn typeof value === \"number\" || value === null || value === undefined;\n}\n\n/**\n * Type predicate to validate string value or null\n *\n * @internal\n */\nfunction isStringOrNull(value: unknown): value is string | null {\n\treturn typeof value === \"string\" || value === null || value === undefined;\n}\n\n/**\n * Type predicate to validate boolean value\n *\n * @internal\n */\nfunction isBoolean(value: unknown): value is boolean {\n\treturn typeof value === \"boolean\" || value === undefined;\n}\n\n/**\n * Convert a File or Blob to Uint8Array\n *\n * Handles both browser File API and server-side Blob-like objects,\n * providing a unified interface for reading binary data.\n *\n * @param file - The File or Blob to convert\n * @returns Promise resolving to the byte array\n * @throws {Error} If the file cannot be read or exceeds size limit\n *\n * @example\n * ```typescript\n * const file = document.getElementById('input').files[0];\n * const bytes = await fileToUint8Array(file);\n * const result = await extractBytes(bytes, 'application/pdf');\n * ```\n */\nexport async function fileToUint8Array(file: File | Blob): Promise<Uint8Array> {\n\ttry {\n\t\tif (file.size > MAX_FILE_SIZE) {\n\t\t\tthrow new Error(\n\t\t\t\t`File size (${file.size} bytes) exceeds maximum (${MAX_FILE_SIZE} bytes). Maximum file size is 512 MB.`,\n\t\t\t);\n\t\t}\n\n\t\tconst arrayBuffer = await file.arrayBuffer();\n\t\treturn new Uint8Array(arrayBuffer);\n\t} catch (error) {\n\t\tthrow new Error(`Failed to read file: ${error instanceof Error ? error.message : String(error)}`);\n\t}\n}\n\n/**\n * Normalize ExtractionConfig for WASM processing\n *\n * Converts TypeScript configuration objects to a WASM-compatible format,\n * handling null values, undefined properties, and nested structures.\n *\n * @param config - The extraction configuration or null\n * @returns Normalized configuration object suitable for WASM\n *\n * @example\n * ```typescript\n * const config: ExtractionConfig = {\n *   ocr: { backend: 'tesseract' },\n *   chunking: { maxChars: 1000 }\n * };\n * const wasmConfig = configToJS(config);\n * ```\n */\nexport function configToJS(config: ExtractionConfig | null): Record<string, unknown> {\n\tif (!config) {\n\t\treturn {};\n\t}\n\n\t// Convert camelCase key to snake_case to match Rust serde field names.\n\tconst toSnakeCase = (str: string): string => str.replace(/[A-Z]/g, (letter) => `_${letter.toLowerCase()}`);\n\n\tconst normalizeValue = (value: unknown): unknown => {\n\t\tif (value === null || value === undefined) {\n\t\t\treturn null;\n\t\t}\n\t\tif (typeof value === \"object\") {\n\t\t\tif (Array.isArray(value)) {\n\t\t\t\treturn value.map(normalizeValue);\n\t\t\t}\n\t\t\tconst obj = value as Record<string, unknown>;\n\t\t\tconst normalized: Record<string, unknown> = {};\n\t\t\tfor (const [key, val] of Object.entries(obj)) {\n\t\t\t\tconst normalizedVal = normalizeValue(val);\n\t\t\t\tif (normalizedVal !== null && normalizedVal !== undefined) {\n\t\t\t\t\tnormalized[toSnakeCase(key)] = normalizedVal;\n\t\t\t\t}\n\t\t\t}\n\t\t\treturn Object.keys(normalized).length > 0 ? normalized : null;\n\t\t}\n\t\treturn value;\n\t};\n\n\tconst normalized: Record<string, unknown> = {};\n\tfor (const [key, value] of Object.entries(config)) {\n\t\tconst normalizedValue = normalizeValue(value);\n\t\tif (normalizedValue !== null && normalizedValue !== undefined) {\n\t\t\tnormalized[toSnakeCase(key)] = normalizedValue;\n\t\t}\n\t}\n\n\treturn normalized;\n}\n\n/**\n * Parse WASM extraction result and convert to TypeScript type\n *\n * Handles conversion of WASM-returned objects to proper ExtractionResult types,\n * including proper array conversions and type assertions for tables, chunks, and images.\n *\n * @param jsValue - The raw WASM result value\n * @returns Properly typed ExtractionResult\n * @throws {Error} If the result structure is invalid\n *\n * @example\n * ```typescript\n * const wasmResult = await wasmExtract(bytes, mimeType, config);\n * const result = jsToExtractionResult(wasmResult);\n * console.log(result.content);\n * ```\n */\nexport function jsToExtractionResult(jsValue: unknown): ExtractionResult {\n\tif (!jsValue || typeof jsValue !== \"object\") {\n\t\tthrow new Error(\"Invalid extraction result: value is not an object\");\n\t}\n\n\tconst result = jsValue as Record<string, unknown>;\n\tconst mimeType =\n\t\ttypeof result.mimeType === \"string\"\n\t\t\t? result.mimeType\n\t\t\t: typeof result.mime_type === \"string\"\n\t\t\t\t? result.mime_type\n\t\t\t\t: null;\n\n\tif (typeof result.content !== \"string\") {\n\t\tthrow new Error(\"Invalid extraction result: missing or invalid content\");\n\t}\n\tif (typeof mimeType !== \"string\") {\n\t\tthrow new Error(\"Invalid extraction result: missing or invalid mimeType\");\n\t}\n\tif (!result.metadata || typeof result.metadata !== \"object\") {\n\t\tthrow new Error(\"Invalid extraction result: missing or invalid metadata\");\n\t}\n\n\tconst tables: Table[] = [];\n\tif (Array.isArray(result.tables)) {\n\t\tfor (const table of result.tables) {\n\t\t\tif (table && typeof table === \"object\") {\n\t\t\t\tconst t = table as Record<string, unknown>;\n\t\t\t\tconst pageNumber =\n\t\t\t\t\ttypeof t.pageNumber === \"number\" ? t.pageNumber : typeof t.page_number === \"number\" ? t.page_number : null;\n\t\t\t\tif (\n\t\t\t\t\tArray.isArray(t.cells) &&\n\t\t\t\t\tt.cells.every((row) => Array.isArray(row) && row.every((cell) => typeof cell === \"string\")) &&\n\t\t\t\t\ttypeof t.markdown === \"string\" &&\n\t\t\t\t\tpageNumber !== null\n\t\t\t\t) {\n\t\t\t\t\ttables.push({\n\t\t\t\t\t\tcells: t.cells as string[][],\n\t\t\t\t\t\tmarkdown: t.markdown,\n\t\t\t\t\t\tpageNumber,\n\t\t\t\t\t});\n\t\t\t\t}\n\t\t\t}\n\t\t}\n\t}\n\n\tconst chunks: Chunk[] | null = Array.isArray(result.chunks)\n\t\t? result.chunks.map((chunk) => {\n\t\t\t\tif (!chunk || typeof chunk !== \"object\") {\n\t\t\t\t\tthrow new Error(\"Invalid chunk structure\");\n\t\t\t\t}\n\t\t\t\tconst c = chunk as Record<string, unknown>;\n\t\t\t\tif (typeof c.content !== \"string\") {\n\t\t\t\t\tthrow new Error(\"Invalid chunk: missing content\");\n\t\t\t\t}\n\t\t\t\tif (!c.metadata || typeof c.metadata !== \"object\") {\n\t\t\t\t\tthrow new Error(\"Invalid chunk: missing metadata\");\n\t\t\t\t}\n\t\t\t\tconst metadata = c.metadata as Record<string, unknown>;\n\n\t\t\t\tlet embedding: number[] | null = null;\n\t\t\t\tif (Array.isArray(c.embedding)) {\n\t\t\t\t\tif (!c.embedding.every((item) => typeof item === \"number\")) {\n\t\t\t\t\t\tthrow new Error(\"Invalid chunk: embedding must contain only numbers\");\n\t\t\t\t\t}\n\t\t\t\t\tembedding = c.embedding;\n\t\t\t\t}\n\n\t\t\t\t// Coerce numeric values - handle BigInt, strings, and numbers\n\t\t\t\tconst coerceToNumber = (value: unknown, fieldName: string): number => {\n\t\t\t\t\tif (typeof value === \"number\") {\n\t\t\t\t\t\treturn value;\n\t\t\t\t\t}\n\t\t\t\t\tif (typeof value === \"bigint\") {\n\t\t\t\t\t\treturn Number(value);\n\t\t\t\t\t}\n\t\t\t\t\tif (typeof value === \"string\") {\n\t\t\t\t\t\tconst parsed = parseInt(value, 10);\n\t\t\t\t\t\tif (Number.isNaN(parsed)) {\n\t\t\t\t\t\t\tthrow new Error(`Invalid chunk metadata: ${fieldName} must be a valid number, got \"${value}\"`);\n\t\t\t\t\t\t}\n\t\t\t\t\t\treturn parsed;\n\t\t\t\t\t}\n\t\t\t\t\tthrow new Error(`Invalid chunk metadata: ${fieldName} must be a number, got ${typeof value}`);\n\t\t\t\t};\n\n\t\t\t\t// The Rust code uses snake_case field names (byte_start, byte_end, etc)\n\t\t\t\t// but TypeScript expects camelCase (charStart, charEnd, etc)\n\t\t\t\t// For now, treat byte offsets as character offsets since the content is UTF-8\n\t\t\t\tconst charStart = coerceToNumber(\n\t\t\t\t\tmetadata.charStart ?? metadata.char_start ?? metadata.byteStart ?? metadata.byte_start,\n\t\t\t\t\t\"charStart\",\n\t\t\t\t);\n\t\t\t\tconst charEnd = coerceToNumber(\n\t\t\t\t\tmetadata.charEnd ?? metadata.char_end ?? metadata.byteEnd ?? metadata.byte_end,\n\t\t\t\t\t\"charEnd\",\n\t\t\t\t);\n\t\t\t\tconst chunkIndex = coerceToNumber(metadata.chunkIndex ?? metadata.chunk_index, \"chunkIndex\");\n\t\t\t\tconst totalChunks = coerceToNumber(metadata.totalChunks ?? metadata.total_chunks, \"totalChunks\");\n\n\t\t\t\tlet tokenCount: number | null = null;\n\t\t\t\tconst tokenCountValue = metadata.tokenCount ?? metadata.token_count;\n\t\t\t\tif (tokenCountValue !== null && tokenCountValue !== undefined) {\n\t\t\t\t\ttokenCount = coerceToNumber(tokenCountValue, \"tokenCount\");\n\t\t\t\t}\n\n\t\t\t\treturn {\n\t\t\t\t\tcontent: c.content,\n\t\t\t\t\tembedding,\n\t\t\t\t\tmetadata: {\n\t\t\t\t\t\tcharStart,\n\t\t\t\t\t\tcharEnd,\n\t\t\t\t\t\ttokenCount,\n\t\t\t\t\t\tchunkIndex,\n\t\t\t\t\t\ttotalChunks,\n\t\t\t\t\t},\n\t\t\t\t};\n\t\t\t})\n\t\t: null;\n\n\tconst images: ExtractedImage[] | null = Array.isArray(result.images)\n\t\t? result.images.map((image) => {\n\t\t\t\tif (!image || typeof image !== \"object\") {\n\t\t\t\t\tthrow new Error(\"Invalid image structure\");\n\t\t\t\t}\n\t\t\t\tconst img = image as Record<string, unknown>;\n\t\t\t\tlet imageData: Uint8Array;\n\t\t\t\tif (img.data instanceof Uint8Array) {\n\t\t\t\t\timageData = img.data;\n\t\t\t\t} else if (Array.isArray(img.data)) {\n\t\t\t\t\timageData = new Uint8Array(img.data as number[]);\n\t\t\t\t} else {\n\t\t\t\t\tthrow new Error(\"Invalid image: data must be Uint8Array or array\");\n\t\t\t\t}\n\t\t\t\tif (typeof img.format !== \"string\") {\n\t\t\t\t\tthrow new Error(\"Invalid image: missing format\");\n\t\t\t\t}\n\n\t\t\t\t// Support both camelCase and snake_case field names (Rust serde uses snake_case)\n\t\t\t\tconst imageIndex = img.imageIndex ?? img.image_index;\n\t\t\t\tconst pageNumber = img.pageNumber ?? img.page_number;\n\t\t\t\tconst bitsPerComponent = img.bitsPerComponent ?? img.bits_per_component;\n\t\t\t\tconst isMask = img.isMask ?? img.is_mask;\n\t\t\t\tconst ocrResult = img.ocrResult ?? img.ocr_result;\n\n\t\t\t\tif (typeof imageIndex !== \"number\") {\n\t\t\t\t\tthrow new Error(\"Invalid image: imageIndex must be a number\");\n\t\t\t\t}\n\t\t\t\tif (!isNumberOrNull(pageNumber)) {\n\t\t\t\t\tthrow new Error(\"Invalid image: pageNumber must be a number or null\");\n\t\t\t\t}\n\t\t\t\tif (!isNumberOrNull(img.width)) {\n\t\t\t\t\tthrow new Error(\"Invalid image: width must be a number or null\");\n\t\t\t\t}\n\t\t\t\tif (!isNumberOrNull(img.height)) {\n\t\t\t\t\tthrow new Error(\"Invalid image: height must be a number or null\");\n\t\t\t\t}\n\t\t\t\tif (!isNumberOrNull(bitsPerComponent)) {\n\t\t\t\t\tthrow new Error(\"Invalid image: bitsPerComponent must be a number or null\");\n\t\t\t\t}\n\n\t\t\t\tif (!isBoolean(isMask)) {\n\t\t\t\t\tthrow new Error(\"Invalid image: isMask must be a boolean\");\n\t\t\t\t}\n\n\t\t\t\tif (!isStringOrNull(img.colorspace)) {\n\t\t\t\t\tthrow new Error(\"Invalid image: colorspace must be a string or null\");\n\t\t\t\t}\n\t\t\t\tif (!isStringOrNull(img.description)) {\n\t\t\t\t\tthrow new Error(\"Invalid image: description must be a string or null\");\n\t\t\t\t}\n\n\t\t\t\treturn {\n\t\t\t\t\tdata: imageData,\n\t\t\t\t\tformat: img.format,\n\t\t\t\t\timageIndex: imageIndex,\n\t\t\t\t\tpageNumber: pageNumber ?? null,\n\t\t\t\t\twidth: (img.width as number) ?? null,\n\t\t\t\t\theight: (img.height as number) ?? null,\n\t\t\t\t\tcolorspace: (img.colorspace as string) ?? null,\n\t\t\t\t\tbitsPerComponent: bitsPerComponent ?? null,\n\t\t\t\t\tisMask: isMask ?? false,\n\t\t\t\t\tdescription: (img.description as string) ?? null,\n\t\t\t\t\tocrResult: ocrResult ? jsToExtractionResult(ocrResult) : null,\n\t\t\t\t};\n\t\t\t})\n\t\t: null;\n\n\tlet detectedLanguages: string[] | null = null;\n\tconst detectedLanguagesRaw = Array.isArray(result.detectedLanguages)\n\t\t? result.detectedLanguages\n\t\t: result.detected_languages;\n\tif (Array.isArray(detectedLanguagesRaw)) {\n\t\tif (!detectedLanguagesRaw.every((lang) => typeof lang === \"string\")) {\n\t\t\tthrow new Error(\"Invalid result: detectedLanguages must contain only strings\");\n\t\t}\n\t\tdetectedLanguages = detectedLanguagesRaw;\n\t}\n\n\tconst extractedKeywords = (result.extractedKeywords ?? result.extracted_keywords ?? null) as\n\t\t| ExtractedKeyword[]\n\t\t| null;\n\tconst qualityScore =\n\t\ttypeof (result.qualityScore ?? result.quality_score) === \"number\"\n\t\t\t? ((result.qualityScore ?? result.quality_score) as number)\n\t\t\t: null;\n\tconst processingWarnings = (result.processingWarnings ?? result.processing_warnings ?? null) as\n\t\t| ProcessingWarning[]\n\t\t| null;\n\tconst elements = (result.elements ?? null) as Element[] | null;\n\tconst ocrElements = (result.ocrElements ?? result.ocr_elements ?? null) as OcrElement[] | null;\n\tconst document = (result.document ?? null) as DocumentStructure | null;\n\tconst pages = (result.pages ?? null) as PageContent[] | null;\n\tconst annotations = (result.annotations ?? null) as PdfAnnotation[] | null;\n\n\treturn {\n\t\tcontent: result.content,\n\t\tmimeType,\n\t\tmetadata: (result.metadata ?? {}) as Metadata,\n\t\ttables,\n\t\tdetectedLanguages,\n\t\tchunks,\n\t\timages,\n\t\tpages,\n\t\textractedKeywords,\n\t\tqualityScore,\n\t\tprocessingWarnings,\n\t\telements,\n\t\tocrElements,\n\t\tdocument,\n\t\tannotations,\n\t};\n}\n\n/**\n * Wrap and format WASM errors with context\n *\n * Converts WASM error messages to JavaScript Error objects with proper context\n * and stack trace information when available.\n *\n * @param error - The error from WASM\n * @param context - Additional context about what operation failed\n * @returns A formatted Error object\n *\n * @internal\n *\n * @example\n * ```typescript\n * try {\n *   await wasmExtract(bytes, mimeType);\n * } catch (error) {\n *   throw wrapWasmError(error, 'extracting document');\n * }\n * ```\n */\nexport function wrapWasmError(error: unknown, context: string): Error {\n\tif (error instanceof Error) {\n\t\treturn new Error(`Error ${context}: ${error.message}`, {\n\t\t\tcause: error,\n\t\t});\n\t}\n\n\tconst message = String(error);\n\treturn new Error(`Error ${context}: ${message}`);\n}\n\n/**\n * Validate that a WASM-returned value conforms to ExtractionResult structure\n *\n * Performs structural validation without full type checking,\n * useful for runtime validation of WASM output.\n *\n * @param value - The value to validate\n * @returns True if value appears to be a valid ExtractionResult\n *\n * @internal\n */\nexport function isValidExtractionResult(value: unknown): value is ExtractionResult {\n\tif (!value || typeof value !== \"object\") {\n\t\treturn false;\n\t}\n\n\tconst obj = value as Record<string, unknown>;\n\treturn (\n\t\ttypeof obj.content === \"string\" &&\n\t\t(typeof obj.mimeType === \"string\" || typeof obj.mime_type === \"string\") &&\n\t\tobj.metadata !== null &&\n\t\ttypeof obj.metadata === \"object\" &&\n\t\tArray.isArray(obj.tables)\n\t);\n}\n"],"mappings":";AAiDA,IAAM,gBAAgB,MAAM,OAAO;AAOnC,SAAS,eAAe,OAAwC;AAC/D,SAAO,OAAO,UAAU,YAAY,UAAU,QAAQ,UAAU;AACjE;AAOA,SAAS,eAAe,OAAwC;AAC/D,SAAO,OAAO,UAAU,YAAY,UAAU,QAAQ,UAAU;AACjE;AAOA,SAAS,UAAU,OAAkC;AACpD,SAAO,OAAO,UAAU,aAAa,UAAU;AAChD;AAmBA,eAAsB,iBAAiB,MAAwC;AAC9E,MAAI;AACH,QAAI,KAAK,OAAO,eAAe;AAC9B,YAAM,IAAI;AAAA,QACT,cAAc,KAAK,IAAI,4BAA4B,aAAa;AAAA,MACjE;AAAA,IACD;AAEA,UAAM,cAAc,MAAM,KAAK,YAAY;AAC3C,WAAO,IAAI,WAAW,WAAW;AAAA,EAClC,SAAS,OAAO;AACf,UAAM,IAAI,MAAM,wBAAwB,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK,CAAC,EAAE;AAAA,EACjG;AACD;AAoBO,SAAS,WAAW,QAA0D;AACpF,MAAI,CAAC,QAAQ;AACZ,WAAO,CAAC;AAAA,EACT;AAGA,QAAM,cAAc,CAAC,QAAwB,IAAI,QAAQ,UAAU,CAAC,WAAW,IAAI,OAAO,YAAY,CAAC,EAAE;AAEzG,QAAM,iBAAiB,CAAC,UAA4B;AACnD,QAAI,UAAU,QAAQ,UAAU,QAAW;AAC1C,aAAO;AAAA,IACR;AACA,QAAI,OAAO,UAAU,UAAU;AAC9B,UAAI,MAAM,QAAQ,KAAK,GAAG;AACzB,eAAO,MAAM,IAAI,cAAc;AAAA,MAChC;AACA,YAAM,MAAM;AACZ,YAAMA,cAAsC,CAAC;AAC7C,iBAAW,CAAC,KAAK,GAAG,KAAK,OAAO,QAAQ,GAAG,GAAG;AAC7C,cAAM,gBAAgB,eAAe,GAAG;AACxC,YAAI,kBAAkB,QAAQ,kBAAkB,QAAW;AAC1D,UAAAA,YAAW,YAAY,GAAG,CAAC,IAAI;AAAA,QAChC;AAAA,MACD;AACA,aAAO,OAAO,KAAKA,WAAU,EAAE,SAAS,IAAIA,cAAa;AAAA,IAC1D;AACA,WAAO;AAAA,EACR;AAEA,QAAM,aAAsC,CAAC;AAC7C,aAAW,CAAC,KAAK,KAAK,KAAK,OAAO,QAAQ,MAAM,GAAG;AAClD,UAAM,kBAAkB,eAAe,KAAK;AAC5C,QAAI,oBAAoB,QAAQ,oBAAoB,QAAW;AAC9D,iBAAW,YAAY,GAAG,CAAC,IAAI;AAAA,IAChC;AAAA,EACD;AAEA,SAAO;AACR;AAmBO,SAAS,qBAAqB,SAAoC;AACxE,MAAI,CAAC,WAAW,OAAO,YAAY,UAAU;AAC5C,UAAM,IAAI,MAAM,mDAAmD;AAAA,EACpE;AAEA,QAAM,SAAS;AACf,QAAM,WACL,OAAO,OAAO,aAAa,WACxB,OAAO,WACP,OAAO,OAAO,cAAc,WAC3B,OAAO,YACP;AAEL,MAAI,OAAO,OAAO,YAAY,UAAU;AACvC,UAAM,IAAI,MAAM,uDAAuD;AAAA,EACxE;AACA,MAAI,OAAO,aAAa,UAAU;AACjC,UAAM,IAAI,MAAM,wDAAwD;AAAA,EACzE;AACA,MAAI,CAAC,OAAO,YAAY,OAAO,OAAO,aAAa,UAAU;AAC5D,UAAM,IAAI,MAAM,wDAAwD;AAAA,EACzE;AAEA,QAAM,SAAkB,CAAC;AACzB,MAAI,MAAM,QAAQ,OAAO,MAAM,GAAG;AACjC,eAAW,SAAS,OAAO,QAAQ;AAClC,UAAI,SAAS,OAAO,UAAU,UAAU;AACvC,cAAM,IAAI;AACV,cAAM,aACL,OAAO,EAAE,eAAe,WAAW,EAAE,aAAa,OAAO,EAAE,gBAAgB,WAAW,EAAE,cAAc;AACvG,YACC,MAAM,QAAQ,EAAE,KAAK,KACrB,EAAE,MAAM,MAAM,CAAC,QAAQ,MAAM,QAAQ,GAAG,KAAK,IAAI,MAAM,CAAC,SAAS,OAAO,SAAS,QAAQ,CAAC,KAC1F,OAAO,EAAE,aAAa,YACtB,eAAe,MACd;AACD,iBAAO,KAAK;AAAA,YACX,OAAO,EAAE;AAAA,YACT,UAAU,EAAE;AAAA,YACZ;AAAA,UACD,CAAC;AAAA,QACF;AAAA,MACD;AAAA,IACD;AAAA,EACD;AAEA,QAAM,SAAyB,MAAM,QAAQ,OAAO,MAAM,IACvD,OAAO,OAAO,IAAI,CAAC,UAAU;AAC7B,QAAI,CAAC,SAAS,OAAO,UAAU,UAAU;AACxC,YAAM,IAAI,MAAM,yBAAyB;AAAA,IAC1C;AACA,UAAM,IAAI;AACV,QAAI,OAAO,EAAE,YAAY,UAAU;AAClC,YAAM,IAAI,MAAM,gCAAgC;AAAA,IACjD;AACA,QAAI,CAAC,EAAE,YAAY,OAAO,EAAE,aAAa,UAAU;AAClD,YAAM,IAAI,MAAM,iCAAiC;AAAA,IAClD;AACA,UAAM,WAAW,EAAE;AAEnB,QAAI,YAA6B;AACjC,QAAI,MAAM,QAAQ,EAAE,SAAS,GAAG;AAC/B,UAAI,CAAC,EAAE,UAAU,MAAM,CAAC,SAAS,OAAO,SAAS,QAAQ,GAAG;AAC3D,cAAM,IAAI,MAAM,oDAAoD;AAAA,MACrE;AACA,kBAAY,EAAE;AAAA,IACf;AAGA,UAAM,iBAAiB,CAAC,OAAgB,cAA8B;AACrE,UAAI,OAAO,UAAU,UAAU;AAC9B,eAAO;AAAA,MACR;AACA,UAAI,OAAO,UAAU,UAAU;AAC9B,eAAO,OAAO,KAAK;AAAA,MACpB;AACA,UAAI,OAAO,UAAU,UAAU;AAC9B,cAAM,SAAS,SAAS,OAAO,EAAE;AACjC,YAAI,OAAO,MAAM,MAAM,GAAG;AACzB,gBAAM,IAAI,MAAM,2BAA2B,SAAS,iCAAiC,KAAK,GAAG;AAAA,QAC9F;AACA,eAAO;AAAA,MACR;AACA,YAAM,IAAI,MAAM,2BAA2B,SAAS,0BAA0B,OAAO,KAAK,EAAE;AAAA,IAC7F;AAKA,UAAM,YAAY;AAAA,MACjB,SAAS,aAAa,SAAS,cAAc,SAAS,aAAa,SAAS;AAAA,MAC5E;AAAA,IACD;AACA,UAAM,UAAU;AAAA,MACf,SAAS,WAAW,SAAS,YAAY,SAAS,WAAW,SAAS;AAAA,MACtE;AAAA,IACD;AACA,UAAM,aAAa,eAAe,SAAS,cAAc,SAAS,aAAa,YAAY;AAC3F,UAAM,cAAc,eAAe,SAAS,eAAe,SAAS,cAAc,aAAa;AAE/F,QAAI,aAA4B;AAChC,UAAM,kBAAkB,SAAS,cAAc,SAAS;AACxD,QAAI,oBAAoB,QAAQ,oBAAoB,QAAW;AAC9D,mBAAa,eAAe,iBAAiB,YAAY;AAAA,IAC1D;AAEA,WAAO;AAAA,MACN,SAAS,EAAE;AAAA,MACX;AAAA,MACA,UAAU;AAAA,QACT;AAAA,QACA;AAAA,QACA;AAAA,QACA;AAAA,QACA;AAAA,MACD;AAAA,IACD;AAAA,EACD,CAAC,IACA;AAEH,QAAM,SAAkC,MAAM,QAAQ,OAAO,MAAM,IAChE,OAAO,OAAO,IAAI,CAAC,UAAU;AAC7B,QAAI,CAAC,SAAS,OAAO,UAAU,UAAU;AACxC,YAAM,IAAI,MAAM,yBAAyB;AAAA,IAC1C;AACA,UAAM,MAAM;AACZ,QAAI;AACJ,QAAI,IAAI,gBAAgB,YAAY;AACnC,kBAAY,IAAI;AAAA,IACjB,WAAW,MAAM,QAAQ,IAAI,IAAI,GAAG;AACnC,kBAAY,IAAI,WAAW,IAAI,IAAgB;AAAA,IAChD,OAAO;AACN,YAAM,IAAI,MAAM,iDAAiD;AAAA,IAClE;AACA,QAAI,OAAO,IAAI,WAAW,UAAU;AACnC,YAAM,IAAI,MAAM,+BAA+B;AAAA,IAChD;AAGA,UAAM,aAAa,IAAI,cAAc,IAAI;AACzC,UAAM,aAAa,IAAI,cAAc,IAAI;AACzC,UAAM,mBAAmB,IAAI,oBAAoB,IAAI;AACrD,UAAM,SAAS,IAAI,UAAU,IAAI;AACjC,UAAM,YAAY,IAAI,aAAa,IAAI;AAEvC,QAAI,OAAO,eAAe,UAAU;AACnC,YAAM,IAAI,MAAM,4CAA4C;AAAA,IAC7D;AACA,QAAI,CAAC,eAAe,UAAU,GAAG;AAChC,YAAM,IAAI,MAAM,oDAAoD;AAAA,IACrE;AACA,QAAI,CAAC,eAAe,IAAI,KAAK,GAAG;AAC/B,YAAM,IAAI,MAAM,+CAA+C;AAAA,IAChE;AACA,QAAI,CAAC,eAAe,IAAI,MAAM,GAAG;AAChC,YAAM,IAAI,MAAM,gDAAgD;AAAA,IACjE;AACA,QAAI,CAAC,eAAe,gBAAgB,GAAG;AACtC,YAAM,IAAI,MAAM,0DAA0D;AAAA,IAC3E;AAEA,QAAI,CAAC,UAAU,MAAM,GAAG;AACvB,YAAM,IAAI,MAAM,yCAAyC;AAAA,IAC1D;AAEA,QAAI,CAAC,eAAe,IAAI,UAAU,GAAG;AACpC,YAAM,IAAI,MAAM,oDAAoD;AAAA,IACrE;AACA,QAAI,CAAC,eAAe,IAAI,WAAW,GAAG;AACrC,YAAM,IAAI,MAAM,qDAAqD;AAAA,IACtE;AAEA,WAAO;AAAA,MACN,MAAM;AAAA,MACN,QAAQ,IAAI;AAAA,MACZ;AAAA,MACA,YAAY,cAAc;AAAA,MAC1B,OAAQ,IAAI,SAAoB;AAAA,MAChC,QAAS,IAAI,UAAqB;AAAA,MAClC,YAAa,IAAI,cAAyB;AAAA,MAC1C,kBAAkB,oBAAoB;AAAA,MACtC,QAAQ,UAAU;AAAA,MAClB,aAAc,IAAI,eAA0B;AAAA,MAC5C,WAAW,YAAY,qBAAqB,SAAS,IAAI;AAAA,IAC1D;AAAA,EACD,CAAC,IACA;AAEH,MAAI,oBAAqC;AACzC,QAAM,uBAAuB,MAAM,QAAQ,OAAO,iBAAiB,IAChE,OAAO,oBACP,OAAO;AACV,MAAI,MAAM,QAAQ,oBAAoB,GAAG;AACxC,QAAI,CAAC,qBAAqB,MAAM,CAAC,SAAS,OAAO,SAAS,QAAQ,GAAG;AACpE,YAAM,IAAI,MAAM,6DAA6D;AAAA,IAC9E;AACA,wBAAoB;AAAA,EACrB;AAEA,QAAM,oBAAqB,OAAO,qBAAqB,OAAO,sBAAsB;AAGpF,QAAM,eACL,QAAQ,OAAO,gBAAgB,OAAO,mBAAmB,WACpD,OAAO,gBAAgB,OAAO,gBAChC;AACJ,QAAM,qBAAsB,OAAO,sBAAsB,OAAO,uBAAuB;AAGvF,QAAM,WAAY,OAAO,YAAY;AACrC,QAAM,cAAe,OAAO,eAAe,OAAO,gBAAgB;AAClE,QAAM,WAAY,OAAO,YAAY;AACrC,QAAM,QAAS,OAAO,SAAS;AAC/B,QAAM,cAAe,OAAO,eAAe;AAE3C,SAAO;AAAA,IACN,SAAS,OAAO;AAAA,IAChB;AAAA,IACA,UAAW,OAAO,YAAY,CAAC;AAAA,IAC/B;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,EACD;AACD;AAuBO,SAAS,cAAc,OAAgB,SAAwB;AACrE,MAAI,iBAAiB,OAAO;AAC3B,WAAO,IAAI,MAAM,SAAS,OAAO,KAAK,MAAM,OAAO,IAAI;AAAA,MACtD,OAAO;AAAA,IACR,CAAC;AAAA,EACF;AAEA,QAAM,UAAU,OAAO,KAAK;AAC5B,SAAO,IAAI,MAAM,SAAS,OAAO,KAAK,OAAO,EAAE;AAChD;AAaO,SAAS,wBAAwB,OAA2C;AAClF,MAAI,CAAC,SAAS,OAAO,UAAU,UAAU;AACxC,WAAO;AAAA,EACR;AAEA,QAAM,MAAM;AACZ,SACC,OAAO,IAAI,YAAY,aACtB,OAAO,IAAI,aAAa,YAAY,OAAO,IAAI,cAAc,aAC9D,IAAI,aAAa,QACjB,OAAO,IAAI,aAAa,YACxB,MAAM,QAAQ,IAAI,MAAM;AAE1B;","names":["normalized"]}

package/dist/index.js

@@ -1,29 +1,3 @@
-var __defProp = Object.defineProperty;
-var __getOwnPropNames = Object.getOwnPropertyNames;
-var __esm = (fn, res) => function __init() {
-  return fn && (res = (0, fn[__getOwnPropNames(fn)[0]])(fn = 0)), res;
-};
-var __export = (target, all) => {
-  for (var name in all)
-    __defProp(target, name, { get: all[name], enumerable: true });
-};
-
-// typescript/pdfium.js
-var pdfium_exports = {};
-__export(pdfium_exports, {
-  default: () => initPdfium
-});
-async function initPdfium() {
-  return {
-    // Dummy implementation for testing
-  };
-}
-var init_pdfium = __esm({
-  "typescript/pdfium.js"() {
-    "use strict";
-  }
-});
-
 // typescript/runtime.ts
 function detectRuntime() {
   const globalCaches = globalThis.caches;
@@ -174,16 +148,72 @@ function getRuntimeInfo() {
 }
 
 // typescript/initialization/pdfium-loader.ts
+async function loadPdfiumForNode() {
+  try {
+    const fs = await import(
+      /* @vite-ignore */
+      "fs/promises"
+    );
+    const path = await import(
+      /* @vite-ignore */
+      "path"
+    );
+    const url = await import(
+      /* @vite-ignore */
+      "url"
+    );
+    const __dirname = path.dirname(url.fileURLToPath(import.meta.url));
+    const envPath = process.env.KREUZBERG_PDFIUM_PATH;
+    const candidates = [];
+    if (envPath) {
+      candidates.push(path.join(envPath, "pdfium.js"));
+      candidates.push(envPath);
+    }
+    candidates.push(
+      path.join(__dirname, "..", "pdfium.js"),
+      // dist/pdfium.js
+      path.join(__dirname, "pdfium.js"),
+      // dist/initialization/pdfium.js
+      path.join(__dirname, "..", "..", "pdfium.js")
+      // package root pdfium.js
+    );
+    for (const candidate of candidates) {
+      try {
+        await fs.access(candidate);
+        const moduleUrl = url.pathToFileURL(candidate).href;
+        return await import(
+          /* @vite-ignore */
+          moduleUrl
+        );
+      } catch {
+      }
+    }
+    return null;
+  } catch {
+    return null;
+  }
+}
+async function loadPdfiumModule() {
+  if (isNode()) {
+    return loadPdfiumForNode();
+  }
+  try {
+    return await import("./pdfium.js");
+  } catch {
+    return null;
+  }
+}
 async function initializePdfiumAsync(wasmModule) {
   if (!wasmModule || typeof wasmModule.initialize_pdfium_render !== "function") {
     return;
   }
-  if (!isBrowser()) {
-    console.debug("PDFium initialization skipped (non-browser environment)");
-    return;
-  }
   try {
-    const pdfiumModule = await Promise.resolve().then(() => (init_pdfium(), pdfium_exports));
+    const pdfiumModule = await loadPdfiumModule();
+    if (!pdfiumModule) {
+      console.debug("PDFium module not found, PDF extraction will not be available");
+      console.debug("To enable PDF support, provide pdfium.js via KREUZBERG_PDFIUM_PATH or manual initialization");
+      return;
+    }
     const pdfium = typeof pdfiumModule.default === "function" ? await pdfiumModule.default() : pdfiumModule;
     const success = wasmModule.initialize_pdfium_render(pdfium, wasmModule, false);
     if (!success) {
@@ -222,7 +252,7 @@ function configToJS(config) {
   if (!config) {
     return {};
   }
-  const normalized = {};
+  const toSnakeCase = (str) => str.replace(/[A-Z]/g, (letter) => `_${letter.toLowerCase()}`);
   const normalizeValue = (value) => {
     if (value === null || value === void 0) {
       return null;
@@ -236,17 +266,18 @@ function configToJS(config) {
       for (const [key, val] of Object.entries(obj)) {
         const normalizedVal = normalizeValue(val);
         if (normalizedVal !== null && normalizedVal !== void 0) {
-          normalized2[key] = normalizedVal;
+          normalized2[toSnakeCase(key)] = normalizedVal;
         }
       }
       return Object.keys(normalized2).length > 0 ? normalized2 : null;
     }
     return value;
   };
+  const normalized = {};
   for (const [key, value] of Object.entries(config)) {
     const normalizedValue = normalizeValue(value);
     if (normalizedValue !== null && normalizedValue !== void 0) {
-      normalized[key] = normalizedValue;
+      normalized[toSnakeCase(key)] = normalizedValue;
     }
   }
   return normalized;
@@ -349,8 +380,13 @@ function jsToExtractionResult(jsValue) {
       throw new Error("Invalid image structure");
     }
     const img = image;
-    if (!(img.data instanceof Uint8Array)) {
-      throw new Error("Invalid image: data must be Uint8Array");
+    let imageData;
+    if (img.data instanceof Uint8Array) {
+      imageData = img.data;
+    } else if (Array.isArray(img.data)) {
+      imageData = new Uint8Array(img.data);
+    } else {
+      throw new Error("Invalid image: data must be Uint8Array or array");
     }
     if (typeof img.format !== "string") {
       throw new Error("Invalid image: missing format");
@@ -385,7 +421,7 @@ function jsToExtractionResult(jsValue) {
       throw new Error("Invalid image: description must be a string or null");
     }
     return {
-      data: img.data,
+      data: imageData,
       format: img.format,
       imageIndex,
       pageNumber: pageNumber ?? null,
@@ -413,6 +449,7 @@ function jsToExtractionResult(jsValue) {
   const ocrElements = result.ocrElements ?? result.ocr_elements ?? null;
   const document2 = result.document ?? null;
   const pages = result.pages ?? null;
+  const annotations = result.annotations ?? null;
   return {
     content: result.content,
     mimeType,
@@ -427,7 +464,8 @@ function jsToExtractionResult(jsValue) {
     processingWarnings,
     elements,
     ocrElements,
-    document: document2
+    document: document2,
+    annotations
   };
 }
 function wrapWasmError(error, context) {
@@ -1261,24 +1299,173 @@ var TesseractWasmBackend = class {
 };
 
 // typescript/ocr/enabler.ts
+var TESSDATA_CDN_BASE = "https://raw.githubusercontent.com/tesseract-ocr/tessdata_fast/main";
+var NativeWasmOcrBackend = class {
+  tessdataCache = /* @__PURE__ */ new Map();
+  tessdataCdnBase = TESSDATA_CDN_BASE;
+  progressCallback = null;
+  name() {
+    return "kreuzberg-tesseract";
+  }
+  supportedLanguages() {
+    return [
+      "eng",
+      "deu",
+      "fra",
+      "spa",
+      "ita",
+      "por",
+      "nld",
+      "rus",
+      "jpn",
+      "kor",
+      "chi_sim",
+      "chi_tra",
+      "pol",
+      "tur",
+      "swe",
+      "dan",
+      "fin",
+      "nor",
+      "ces",
+      "slk",
+      "ron",
+      "hun",
+      "hrv",
+      "srp",
+      "bul",
+      "ukr",
+      "ell",
+      "ara",
+      "heb",
+      "hin",
+      "tha",
+      "vie",
+      "mkd",
+      "ben",
+      "tam",
+      "tel",
+      "kan",
+      "mal",
+      "mya",
+      "khm",
+      "lao",
+      "sin"
+    ];
+  }
+  async initialize() {
+    const wasm2 = getWasmModule();
+    if (!wasm2?.ocrIsAvailable || !wasm2.ocrIsAvailable()) {
+      throw new Error(
+        "Native WASM OCR is not available. Build with the 'ocr-wasm' feature to enable kreuzberg-tesseract."
+      );
+    }
+  }
+  async shutdown() {
+    this.tessdataCache.clear();
+    this.progressCallback = null;
+  }
+  setProgressCallback(callback) {
+    this.progressCallback = callback;
+  }
+  async processImage(imageBytes, language) {
+    const wasm2 = getWasmModule();
+    if (!wasm2?.ocrRecognize) {
+      throw new Error("Native WASM OCR function not available");
+    }
+    const normalizedLang = language.toLowerCase();
+    this.reportProgress(10);
+    const tessdata = await this.getTessdata(normalizedLang);
+    this.reportProgress(40);
+    let imageData;
+    if (typeof imageBytes === "string") {
+      const binaryString = atob(imageBytes);
+      imageData = new Uint8Array(binaryString.length);
+      for (let i = 0; i < binaryString.length; i++) {
+        imageData[i] = binaryString.charCodeAt(i);
+      }
+    } else {
+      imageData = imageBytes;
+    }
+    this.reportProgress(50);
+    const text = wasm2.ocrRecognize(imageData, tessdata, normalizedLang);
+    this.reportProgress(90);
+    return {
+      content: text,
+      mime_type: "text/plain",
+      metadata: { language: normalizedLang },
+      tables: []
+    };
+  }
+  async getTessdata(language) {
+    const cached = this.tessdataCache.get(language);
+    if (cached) {
+      return cached;
+    }
+    const url = `${this.tessdataCdnBase}/${language}.traineddata`;
+    const response = await fetch(url);
+    if (!response.ok) {
+      throw new Error(`Failed to download tessdata for "${language}" from ${url}: ${response.status}`);
+    }
+    const data = new Uint8Array(await response.arrayBuffer());
+    this.tessdataCache.set(language, data);
+    return data;
+  }
+  reportProgress(progress) {
+    if (this.progressCallback) {
+      try {
+        this.progressCallback(Math.min(100, Math.max(0, progress)));
+      } catch {
+      }
+    }
+  }
+};
 async function enableOcr() {
   if (!isInitialized2()) {
     throw new Error("WASM module not initialized. Call initWasm() first.");
   }
-  if (!isBrowser()) {
+  try {
+    const wasm2 = getWasmModule();
+    if (wasm2?.ocrIsAvailable?.()) {
+      const backend = new NativeWasmOcrBackend();
+      await backend.initialize();
+      registerOcrBackend(backend);
+      registerBackendInRustRegistry(wasm2, backend);
+      return;
+    }
+    if (isBrowser()) {
+      const backend = new TesseractWasmBackend();
+      await backend.initialize();
+      registerOcrBackend(backend);
+      registerBackendInRustRegistry(wasm2, backend);
+      return;
+    }
     throw new Error(
-      "OCR is only available in browser environments. TesseractWasmBackend requires Web Workers and createImageBitmap."
+      "No OCR backend available. Build with the 'ocr-wasm' feature to enable native Tesseract OCR in all environments, or use a browser environment with the tesseract-wasm npm package."
     );
-  }
-  try {
-    const backend = new TesseractWasmBackend();
-    await backend.initialize();
-    registerOcrBackend(backend);
   } catch (error) {
     const message = error instanceof Error ? error.message : String(error);
     throw new Error(`Failed to enable OCR: ${message}`);
   }
 }
+function registerBackendInRustRegistry(wasm2, backend) {
+  const registerFn = wasm2?.register_ocr_backend;
+  if (!registerFn) {
+    return;
+  }
+  const rustAdapter = {
+    name: () => "tesseract",
+    supportedLanguages: () => backend.supportedLanguages?.() ?? ["eng"],
+    processImage: async (imageBase64, language) => {
+      const result = await backend.processImage(imageBase64, language);
+      return typeof result === "string" ? result : JSON.stringify(result);
+    }
+  };
+  try {
+    registerFn(rustAdapter);
+  } catch {
+  }
+}
 
 // typescript/plugin-registry.ts
 var postProcessors = /* @__PURE__ */ new Map();