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

package/dist/adapters/wasm-adapter.cjs.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 { Chunk, ExtractedImage, ExtractionConfig, ExtractionResult, Metadata, Table } 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;\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;\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\";\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\t// Recursively normalize nested objects\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\t// Validate required fields\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\t// Parse tables\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\t// Validate table structure before type casting\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\ttypeof t.pageNumber === \"number\"\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: t.pageNumber,\n\t\t\t\t\t});\n\t\t\t\t}\n\t\t\t}\n\t\t}\n\t}\n\n\t// Parse chunks\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\t// Validate embedding array contains only numbers\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// Validate metadata fields\n\t\t\t\tif (typeof metadata.charStart !== \"number\") {\n\t\t\t\t\tthrow new Error(\"Invalid chunk metadata: charStart must be a number\");\n\t\t\t\t}\n\t\t\t\tif (typeof metadata.charEnd !== \"number\") {\n\t\t\t\t\tthrow new Error(\"Invalid chunk metadata: charEnd must be a number\");\n\t\t\t\t}\n\t\t\t\tif (!isNumberOrNull(metadata.tokenCount)) {\n\t\t\t\t\tthrow new Error(\"Invalid chunk metadata: tokenCount must be a number or null\");\n\t\t\t\t}\n\t\t\t\tif (typeof metadata.chunkIndex !== \"number\") {\n\t\t\t\t\tthrow new Error(\"Invalid chunk metadata: chunkIndex must be a number\");\n\t\t\t\t}\n\t\t\t\tif (typeof metadata.totalChunks !== \"number\") {\n\t\t\t\t\tthrow new Error(\"Invalid chunk metadata: totalChunks must be a number\");\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: metadata.charStart,\n\t\t\t\t\t\tcharEnd: metadata.charEnd,\n\t\t\t\t\t\ttokenCount: metadata.tokenCount,\n\t\t\t\t\t\tchunkIndex: metadata.chunkIndex,\n\t\t\t\t\t\ttotalChunks: metadata.totalChunks,\n\t\t\t\t\t},\n\t\t\t\t};\n\t\t\t})\n\t\t: null;\n\n\t// Parse images\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// Validate numeric fields\n\t\t\t\tif (typeof img.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(img.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(img.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\t// Validate boolean field\n\t\t\t\tif (!isBoolean(img.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\t// Validate string fields\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: img.imageIndex,\n\t\t\t\t\tpageNumber: img.pageNumber,\n\t\t\t\t\twidth: img.width,\n\t\t\t\t\theight: img.height,\n\t\t\t\t\tcolorspace: img.colorspace,\n\t\t\t\t\tbitsPerComponent: img.bitsPerComponent,\n\t\t\t\t\tisMask: img.isMask,\n\t\t\t\t\tdescription: img.description,\n\t\t\t\t\tocrResult: img.ocrResult ? jsToExtractionResult(img.ocrResult) : null,\n\t\t\t\t};\n\t\t\t})\n\t\t: null;\n\n\t// Validate detectedLanguages array\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\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};\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":";;;;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAmCA,IAAM,gBAAgB,MAAM,OAAO;AAOnC,SAAS,eAAe,OAAwC;AAC/D,SAAO,OAAO,UAAU,YAAY,UAAU;AAC/C;AAOA,SAAS,eAAe,OAAwC;AAC/D,SAAO,OAAO,UAAU,YAAY,UAAU;AAC/C;AAOA,SAAS,UAAU,OAAkC;AACpD,SAAO,OAAO,UAAU;AACzB;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;AAG7C,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;AAGL,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;AAGA,QAAM,SAAkB,CAAC;AACzB,MAAI,MAAM,QAAQ,OAAO,MAAM,GAAG;AACjC,eAAW,SAAS,OAAO,QAAQ;AAClC,UAAI,SAAS,OAAO,UAAU,UAAU;AACvC,cAAM,IAAI;AAEV,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,OAAO,EAAE,eAAe,UACvB;AACD,iBAAO,KAAK;AAAA,YACX,OAAO,EAAE;AAAA,YACT,UAAU,EAAE;AAAA,YACZ,YAAY,EAAE;AAAA,UACf,CAAC;AAAA,QACF;AAAA,MACD;AAAA,IACD;AAAA,EACD;AAGA,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;AAGnB,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,QAAI,OAAO,SAAS,cAAc,UAAU;AAC3C,YAAM,IAAI,MAAM,oDAAoD;AAAA,IACrE;AACA,QAAI,OAAO,SAAS,YAAY,UAAU;AACzC,YAAM,IAAI,MAAM,kDAAkD;AAAA,IACnE;AACA,QAAI,CAAC,eAAe,SAAS,UAAU,GAAG;AACzC,YAAM,IAAI,MAAM,6DAA6D;AAAA,IAC9E;AACA,QAAI,OAAO,SAAS,eAAe,UAAU;AAC5C,YAAM,IAAI,MAAM,qDAAqD;AAAA,IACtE;AACA,QAAI,OAAO,SAAS,gBAAgB,UAAU;AAC7C,YAAM,IAAI,MAAM,sDAAsD;AAAA,IACvE;AAEA,WAAO;AAAA,MACN,SAAS,EAAE;AAAA,MACX;AAAA,MACA,UAAU;AAAA,QACT,WAAW,SAAS;AAAA,QACpB,SAAS,SAAS;AAAA,QAClB,YAAY,SAAS;AAAA,QACrB,YAAY,SAAS;AAAA,QACrB,aAAa,SAAS;AAAA,MACvB;AAAA,IACD;AAAA,EACD,CAAC,IACA;AAGH,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,QAAI,OAAO,IAAI,eAAe,UAAU;AACvC,YAAM,IAAI,MAAM,4CAA4C;AAAA,IAC7D;AACA,QAAI,CAAC,eAAe,IAAI,UAAU,GAAG;AACpC,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,IAAI,gBAAgB,GAAG;AAC1C,YAAM,IAAI,MAAM,0DAA0D;AAAA,IAC3E;AAGA,QAAI,CAAC,UAAU,IAAI,MAAM,GAAG;AAC3B,YAAM,IAAI,MAAM,yCAAyC;AAAA,IAC1D;AAGA,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,YAAY,IAAI;AAAA,MAChB,YAAY,IAAI;AAAA,MAChB,OAAO,IAAI;AAAA,MACX,QAAQ,IAAI;AAAA,MACZ,YAAY,IAAI;AAAA,MAChB,kBAAkB,IAAI;AAAA,MACtB,QAAQ,IAAI;AAAA,MACZ,aAAa,IAAI;AAAA,MACjB,WAAW,IAAI,YAAY,qBAAqB,IAAI,SAAS,IAAI;AAAA,IAClE;AAAA,EACD,CAAC,IACA;AAGH,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,SAAO;AAAA,IACN,SAAS,OAAO;AAAA,IAChB;AAAA,IACA,UAAW,OAAO,YAAY,CAAC;AAAA,IAC/B;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 { Chunk, ExtractedImage, ExtractionConfig, ExtractionResult, Metadata, Table } 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;\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;\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\";\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\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\ttypeof t.pageNumber === \"number\"\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: t.pageNumber,\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\tif (typeof metadata.charStart !== \"number\") {\n\t\t\t\t\tthrow new Error(\"Invalid chunk metadata: charStart must be a number\");\n\t\t\t\t}\n\t\t\t\tif (typeof metadata.charEnd !== \"number\") {\n\t\t\t\t\tthrow new Error(\"Invalid chunk metadata: charEnd must be a number\");\n\t\t\t\t}\n\t\t\t\tif (!isNumberOrNull(metadata.tokenCount)) {\n\t\t\t\t\tthrow new Error(\"Invalid chunk metadata: tokenCount must be a number or null\");\n\t\t\t\t}\n\t\t\t\tif (typeof metadata.chunkIndex !== \"number\") {\n\t\t\t\t\tthrow new Error(\"Invalid chunk metadata: chunkIndex must be a number\");\n\t\t\t\t}\n\t\t\t\tif (typeof metadata.totalChunks !== \"number\") {\n\t\t\t\t\tthrow new Error(\"Invalid chunk metadata: totalChunks must be a number\");\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: metadata.charStart,\n\t\t\t\t\t\tcharEnd: metadata.charEnd,\n\t\t\t\t\t\ttokenCount: metadata.tokenCount,\n\t\t\t\t\t\tchunkIndex: metadata.chunkIndex,\n\t\t\t\t\t\ttotalChunks: metadata.totalChunks,\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\tif (typeof img.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(img.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(img.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(img.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: img.imageIndex,\n\t\t\t\t\tpageNumber: img.pageNumber,\n\t\t\t\t\twidth: img.width,\n\t\t\t\t\theight: img.height,\n\t\t\t\t\tcolorspace: img.colorspace,\n\t\t\t\t\tbitsPerComponent: img.bitsPerComponent,\n\t\t\t\t\tisMask: img.isMask,\n\t\t\t\t\tdescription: img.description,\n\t\t\t\t\tocrResult: img.ocrResult ? jsToExtractionResult(img.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\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};\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":";;;;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAmCA,IAAM,gBAAgB,MAAM,OAAO;AAOnC,SAAS,eAAe,OAAwC;AAC/D,SAAO,OAAO,UAAU,YAAY,UAAU;AAC/C;AAOA,SAAS,eAAe,OAAwC;AAC/D,SAAO,OAAO,UAAU,YAAY,UAAU;AAC/C;AAOA,SAAS,UAAU,OAAkC;AACpD,SAAO,OAAO,UAAU;AACzB;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,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,OAAO,EAAE,eAAe,UACvB;AACD,iBAAO,KAAK;AAAA,YACX,OAAO,EAAE;AAAA,YACT,UAAU,EAAE;AAAA,YACZ,YAAY,EAAE;AAAA,UACf,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;AAEA,QAAI,OAAO,SAAS,cAAc,UAAU;AAC3C,YAAM,IAAI,MAAM,oDAAoD;AAAA,IACrE;AACA,QAAI,OAAO,SAAS,YAAY,UAAU;AACzC,YAAM,IAAI,MAAM,kDAAkD;AAAA,IACnE;AACA,QAAI,CAAC,eAAe,SAAS,UAAU,GAAG;AACzC,YAAM,IAAI,MAAM,6DAA6D;AAAA,IAC9E;AACA,QAAI,OAAO,SAAS,eAAe,UAAU;AAC5C,YAAM,IAAI,MAAM,qDAAqD;AAAA,IACtE;AACA,QAAI,OAAO,SAAS,gBAAgB,UAAU;AAC7C,YAAM,IAAI,MAAM,sDAAsD;AAAA,IACvE;AAEA,WAAO;AAAA,MACN,SAAS,EAAE;AAAA,MACX;AAAA,MACA,UAAU;AAAA,QACT,WAAW,SAAS;AAAA,QACpB,SAAS,SAAS;AAAA,QAClB,YAAY,SAAS;AAAA,QACrB,YAAY,SAAS;AAAA,QACrB,aAAa,SAAS;AAAA,MACvB;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;AAEA,QAAI,OAAO,IAAI,eAAe,UAAU;AACvC,YAAM,IAAI,MAAM,4CAA4C;AAAA,IAC7D;AACA,QAAI,CAAC,eAAe,IAAI,UAAU,GAAG;AACpC,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,IAAI,gBAAgB,GAAG;AAC1C,YAAM,IAAI,MAAM,0DAA0D;AAAA,IAC3E;AAEA,QAAI,CAAC,UAAU,IAAI,MAAM,GAAG;AAC3B,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,YAAY,IAAI;AAAA,MAChB,YAAY,IAAI;AAAA,MAChB,OAAO,IAAI;AAAA,MACX,QAAQ,IAAI;AAAA,MACZ,YAAY,IAAI;AAAA,MAChB,kBAAkB,IAAI;AAAA,MACtB,QAAQ,IAAI;AAAA,MACZ,aAAa,IAAI;AAAA,MACjB,WAAW,IAAI,YAAY,qBAAqB,IAAI,SAAS,IAAI;AAAA,IAClE;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,SAAO;AAAA,IACN,SAAS,OAAO;AAAA,IAChB;AAAA,IACA,UAAW,OAAO,YAAY,CAAC;AAAA,IAC/B;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/adapters/wasm-adapter.d.cts

@@ -1,4 +1,4 @@
-import { E as ExtractionConfig, a as ExtractionResult } from '../types-CKjcIYcX.cjs';
+import { a as ExtractionConfig, E as ExtractionResult } from '../types-wVLLDHkl.cjs';
 
 /**
  * WASM Type Adapter

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

@@ -1,4 +1,4 @@
-import { E as ExtractionConfig, a as ExtractionResult } from '../types-CKjcIYcX.js';
+import { a as ExtractionConfig, E as ExtractionResult } from '../types-wVLLDHkl.js';
 
 /**
  * WASM Type Adapter

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 { Chunk, ExtractedImage, ExtractionConfig, ExtractionResult, Metadata, Table } 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;\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;\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\";\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\t// Recursively normalize nested objects\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\t// Validate required fields\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\t// Parse tables\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\t// Validate table structure before type casting\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\ttypeof t.pageNumber === \"number\"\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: t.pageNumber,\n\t\t\t\t\t});\n\t\t\t\t}\n\t\t\t}\n\t\t}\n\t}\n\n\t// Parse chunks\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\t// Validate embedding array contains only numbers\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// Validate metadata fields\n\t\t\t\tif (typeof metadata.charStart !== \"number\") {\n\t\t\t\t\tthrow new Error(\"Invalid chunk metadata: charStart must be a number\");\n\t\t\t\t}\n\t\t\t\tif (typeof metadata.charEnd !== \"number\") {\n\t\t\t\t\tthrow new Error(\"Invalid chunk metadata: charEnd must be a number\");\n\t\t\t\t}\n\t\t\t\tif (!isNumberOrNull(metadata.tokenCount)) {\n\t\t\t\t\tthrow new Error(\"Invalid chunk metadata: tokenCount must be a number or null\");\n\t\t\t\t}\n\t\t\t\tif (typeof metadata.chunkIndex !== \"number\") {\n\t\t\t\t\tthrow new Error(\"Invalid chunk metadata: chunkIndex must be a number\");\n\t\t\t\t}\n\t\t\t\tif (typeof metadata.totalChunks !== \"number\") {\n\t\t\t\t\tthrow new Error(\"Invalid chunk metadata: totalChunks must be a number\");\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: metadata.charStart,\n\t\t\t\t\t\tcharEnd: metadata.charEnd,\n\t\t\t\t\t\ttokenCount: metadata.tokenCount,\n\t\t\t\t\t\tchunkIndex: metadata.chunkIndex,\n\t\t\t\t\t\ttotalChunks: metadata.totalChunks,\n\t\t\t\t\t},\n\t\t\t\t};\n\t\t\t})\n\t\t: null;\n\n\t// Parse images\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// Validate numeric fields\n\t\t\t\tif (typeof img.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(img.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(img.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\t// Validate boolean field\n\t\t\t\tif (!isBoolean(img.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\t// Validate string fields\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: img.imageIndex,\n\t\t\t\t\tpageNumber: img.pageNumber,\n\t\t\t\t\twidth: img.width,\n\t\t\t\t\theight: img.height,\n\t\t\t\t\tcolorspace: img.colorspace,\n\t\t\t\t\tbitsPerComponent: img.bitsPerComponent,\n\t\t\t\t\tisMask: img.isMask,\n\t\t\t\t\tdescription: img.description,\n\t\t\t\t\tocrResult: img.ocrResult ? jsToExtractionResult(img.ocrResult) : null,\n\t\t\t\t};\n\t\t\t})\n\t\t: null;\n\n\t// Validate detectedLanguages array\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\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};\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":";AAmCA,IAAM,gBAAgB,MAAM,OAAO;AAOnC,SAAS,eAAe,OAAwC;AAC/D,SAAO,OAAO,UAAU,YAAY,UAAU;AAC/C;AAOA,SAAS,eAAe,OAAwC;AAC/D,SAAO,OAAO,UAAU,YAAY,UAAU;AAC/C;AAOA,SAAS,UAAU,OAAkC;AACpD,SAAO,OAAO,UAAU;AACzB;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;AAG7C,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;AAGL,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;AAGA,QAAM,SAAkB,CAAC;AACzB,MAAI,MAAM,QAAQ,OAAO,MAAM,GAAG;AACjC,eAAW,SAAS,OAAO,QAAQ;AAClC,UAAI,SAAS,OAAO,UAAU,UAAU;AACvC,cAAM,IAAI;AAEV,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,OAAO,EAAE,eAAe,UACvB;AACD,iBAAO,KAAK;AAAA,YACX,OAAO,EAAE;AAAA,YACT,UAAU,EAAE;AAAA,YACZ,YAAY,EAAE;AAAA,UACf,CAAC;AAAA,QACF;AAAA,MACD;AAAA,IACD;AAAA,EACD;AAGA,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;AAGnB,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,QAAI,OAAO,SAAS,cAAc,UAAU;AAC3C,YAAM,IAAI,MAAM,oDAAoD;AAAA,IACrE;AACA,QAAI,OAAO,SAAS,YAAY,UAAU;AACzC,YAAM,IAAI,MAAM,kDAAkD;AAAA,IACnE;AACA,QAAI,CAAC,eAAe,SAAS,UAAU,GAAG;AACzC,YAAM,IAAI,MAAM,6DAA6D;AAAA,IAC9E;AACA,QAAI,OAAO,SAAS,eAAe,UAAU;AAC5C,YAAM,IAAI,MAAM,qDAAqD;AAAA,IACtE;AACA,QAAI,OAAO,SAAS,gBAAgB,UAAU;AAC7C,YAAM,IAAI,MAAM,sDAAsD;AAAA,IACvE;AAEA,WAAO;AAAA,MACN,SAAS,EAAE;AAAA,MACX;AAAA,MACA,UAAU;AAAA,QACT,WAAW,SAAS;AAAA,QACpB,SAAS,SAAS;AAAA,QAClB,YAAY,SAAS;AAAA,QACrB,YAAY,SAAS;AAAA,QACrB,aAAa,SAAS;AAAA,MACvB;AAAA,IACD;AAAA,EACD,CAAC,IACA;AAGH,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,QAAI,OAAO,IAAI,eAAe,UAAU;AACvC,YAAM,IAAI,MAAM,4CAA4C;AAAA,IAC7D;AACA,QAAI,CAAC,eAAe,IAAI,UAAU,GAAG;AACpC,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,IAAI,gBAAgB,GAAG;AAC1C,YAAM,IAAI,MAAM,0DAA0D;AAAA,IAC3E;AAGA,QAAI,CAAC,UAAU,IAAI,MAAM,GAAG;AAC3B,YAAM,IAAI,MAAM,yCAAyC;AAAA,IAC1D;AAGA,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,YAAY,IAAI;AAAA,MAChB,YAAY,IAAI;AAAA,MAChB,OAAO,IAAI;AAAA,MACX,QAAQ,IAAI;AAAA,MACZ,YAAY,IAAI;AAAA,MAChB,kBAAkB,IAAI;AAAA,MACtB,QAAQ,IAAI;AAAA,MACZ,aAAa,IAAI;AAAA,MACjB,WAAW,IAAI,YAAY,qBAAqB,IAAI,SAAS,IAAI;AAAA,IAClE;AAAA,EACD,CAAC,IACA;AAGH,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,SAAO;AAAA,IACN,SAAS,OAAO;AAAA,IAChB;AAAA,IACA,UAAW,OAAO,YAAY,CAAC;AAAA,IAC/B;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 { Chunk, ExtractedImage, ExtractionConfig, ExtractionResult, Metadata, Table } 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;\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;\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\";\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\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\ttypeof t.pageNumber === \"number\"\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: t.pageNumber,\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\tif (typeof metadata.charStart !== \"number\") {\n\t\t\t\t\tthrow new Error(\"Invalid chunk metadata: charStart must be a number\");\n\t\t\t\t}\n\t\t\t\tif (typeof metadata.charEnd !== \"number\") {\n\t\t\t\t\tthrow new Error(\"Invalid chunk metadata: charEnd must be a number\");\n\t\t\t\t}\n\t\t\t\tif (!isNumberOrNull(metadata.tokenCount)) {\n\t\t\t\t\tthrow new Error(\"Invalid chunk metadata: tokenCount must be a number or null\");\n\t\t\t\t}\n\t\t\t\tif (typeof metadata.chunkIndex !== \"number\") {\n\t\t\t\t\tthrow new Error(\"Invalid chunk metadata: chunkIndex must be a number\");\n\t\t\t\t}\n\t\t\t\tif (typeof metadata.totalChunks !== \"number\") {\n\t\t\t\t\tthrow new Error(\"Invalid chunk metadata: totalChunks must be a number\");\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: metadata.charStart,\n\t\t\t\t\t\tcharEnd: metadata.charEnd,\n\t\t\t\t\t\ttokenCount: metadata.tokenCount,\n\t\t\t\t\t\tchunkIndex: metadata.chunkIndex,\n\t\t\t\t\t\ttotalChunks: metadata.totalChunks,\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\tif (typeof img.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(img.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(img.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(img.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: img.imageIndex,\n\t\t\t\t\tpageNumber: img.pageNumber,\n\t\t\t\t\twidth: img.width,\n\t\t\t\t\theight: img.height,\n\t\t\t\t\tcolorspace: img.colorspace,\n\t\t\t\t\tbitsPerComponent: img.bitsPerComponent,\n\t\t\t\t\tisMask: img.isMask,\n\t\t\t\t\tdescription: img.description,\n\t\t\t\t\tocrResult: img.ocrResult ? jsToExtractionResult(img.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\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};\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":";AAmCA,IAAM,gBAAgB,MAAM,OAAO;AAOnC,SAAS,eAAe,OAAwC;AAC/D,SAAO,OAAO,UAAU,YAAY,UAAU;AAC/C;AAOA,SAAS,eAAe,OAAwC;AAC/D,SAAO,OAAO,UAAU,YAAY,UAAU;AAC/C;AAOA,SAAS,UAAU,OAAkC;AACpD,SAAO,OAAO,UAAU;AACzB;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,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,OAAO,EAAE,eAAe,UACvB;AACD,iBAAO,KAAK;AAAA,YACX,OAAO,EAAE;AAAA,YACT,UAAU,EAAE;AAAA,YACZ,YAAY,EAAE;AAAA,UACf,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;AAEA,QAAI,OAAO,SAAS,cAAc,UAAU;AAC3C,YAAM,IAAI,MAAM,oDAAoD;AAAA,IACrE;AACA,QAAI,OAAO,SAAS,YAAY,UAAU;AACzC,YAAM,IAAI,MAAM,kDAAkD;AAAA,IACnE;AACA,QAAI,CAAC,eAAe,SAAS,UAAU,GAAG;AACzC,YAAM,IAAI,MAAM,6DAA6D;AAAA,IAC9E;AACA,QAAI,OAAO,SAAS,eAAe,UAAU;AAC5C,YAAM,IAAI,MAAM,qDAAqD;AAAA,IACtE;AACA,QAAI,OAAO,SAAS,gBAAgB,UAAU;AAC7C,YAAM,IAAI,MAAM,sDAAsD;AAAA,IACvE;AAEA,WAAO;AAAA,MACN,SAAS,EAAE;AAAA,MACX;AAAA,MACA,UAAU;AAAA,QACT,WAAW,SAAS;AAAA,QACpB,SAAS,SAAS;AAAA,QAClB,YAAY,SAAS;AAAA,QACrB,YAAY,SAAS;AAAA,QACrB,aAAa,SAAS;AAAA,MACvB;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;AAEA,QAAI,OAAO,IAAI,eAAe,UAAU;AACvC,YAAM,IAAI,MAAM,4CAA4C;AAAA,IAC7D;AACA,QAAI,CAAC,eAAe,IAAI,UAAU,GAAG;AACpC,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,IAAI,gBAAgB,GAAG;AAC1C,YAAM,IAAI,MAAM,0DAA0D;AAAA,IAC3E;AAEA,QAAI,CAAC,UAAU,IAAI,MAAM,GAAG;AAC3B,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,YAAY,IAAI;AAAA,MAChB,YAAY,IAAI;AAAA,MAChB,OAAO,IAAI;AAAA,MACX,QAAQ,IAAI;AAAA,MACZ,YAAY,IAAI;AAAA,MAChB,kBAAkB,IAAI;AAAA,MACtB,QAAQ,IAAI;AAAA,MACZ,aAAa,IAAI;AAAA,MACjB,WAAW,IAAI,YAAY,qBAAqB,IAAI,SAAS,IAAI;AAAA,IAClE;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,SAAO;AAAA,IACN,SAAS,OAAO;AAAA,IAChB;AAAA,IACA,UAAW,OAAO,YAAY,CAAC;AAAA,IAC/B;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.cjs

@@ -5,6 +5,9 @@ var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
 var __getOwnPropNames = Object.getOwnPropertyNames;
 var __getProtoOf = Object.getPrototypeOf;
 var __hasOwnProp = Object.prototype.hasOwnProperty;
+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 });
@@ -27,6 +30,22 @@ var __toESM = (mod, isNodeMode, target) => (target = mod != null ? __create(__ge
 ));
 var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
 
+// 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/index.ts
 var index_exports = {};
 __export(index_exports, {
@@ -35,6 +54,8 @@ __export(index_exports, {
   batchExtractBytesSync: () => batchExtractBytesSync,
   batchExtractFiles: () => batchExtractFiles,
   clearOcrBackends: () => clearOcrBackends,
+  clearPostProcessors: () => clearPostProcessors,
+  clearValidators: () => clearValidators,
   configToJS: () => configToJS,
   detectRuntime: () => detectRuntime,
   enableOcr: () => enableOcr,
@@ -45,8 +66,10 @@ __export(index_exports, {
   fileToUint8Array: () => fileToUint8Array,
   getInitializationError: () => getInitializationError,
   getOcrBackend: () => getOcrBackend,
+  getPostProcessor: () => getPostProcessor,
   getRuntimeInfo: () => getRuntimeInfo,
   getRuntimeVersion: () => getRuntimeVersion,
+  getValidator: () => getValidator,
   getVersion: () => getVersion,
   getWasmCapabilities: () => getWasmCapabilities,
   hasBigInt: () => hasBigInt,
@@ -68,8 +91,14 @@ __export(index_exports, {
   isWebEnvironment: () => isWebEnvironment,
   jsToExtractionResult: () => jsToExtractionResult,
   listOcrBackends: () => listOcrBackends,
+  listPostProcessors: () => listPostProcessors,
+  listValidators: () => listValidators,
   registerOcrBackend: () => registerOcrBackend,
+  registerPostProcessor: () => registerPostProcessor,
+  registerValidator: () => registerValidator,
   unregisterOcrBackend: () => unregisterOcrBackend,
+  unregisterPostProcessor: () => unregisterPostProcessor,
+  unregisterValidator: () => unregisterValidator,
   wrapWasmError: () => wrapWasmError
 });
 module.exports = __toCommonJS(index_exports);
@@ -388,93 +417,48 @@ var TesseractWasmBackend = class {
       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;
   }
@@ -576,7 +560,6 @@ var TesseractWasmBackend = class {
           ...pageMetadata
         },
         tables: []
-        // Tesseract-wasm doesn't provide structured table detection
       };
     } catch (error) {
       const message = error instanceof Error ? error.message : String(error);
@@ -862,7 +845,6 @@ function getRuntimeVersion() {
   switch (runtime) {
     case "node":
       return process.version?.substring(1);
-    // Remove 'v' prefix
     case "deno": {
       const deno = globalThis.Deno;
       const version = deno?.version;
@@ -910,6 +892,168 @@ function getRuntimeInfo() {
   };
 }
 
+// typescript/plugin-registry.ts
+var postProcessors = /* @__PURE__ */ new Map();
+var validators = /* @__PURE__ */ new Map();
+function validatePostProcessor(processor) {
+  if (processor === null || processor === void 0) {
+    throw new Error("Post-processor cannot be null or undefined");
+  }
+  const obj = processor;
+  if (typeof obj.name !== "function") {
+    throw new Error("Post-processor must implement name() method");
+  }
+  if (typeof obj.process !== "function") {
+    throw new Error("Post-processor must implement process() method");
+  }
+  const name = obj.name();
+  if (typeof name !== "string" || name.trim() === "") {
+    throw new Error("Post-processor name must be a non-empty string");
+  }
+  return true;
+}
+function registerPostProcessor(processor) {
+  validatePostProcessor(processor);
+  const name = processor.name();
+  if (postProcessors.has(name)) {
+    console.warn(`Post-processor "${name}" already registered, overwriting with new implementation`);
+  }
+  postProcessors.set(name, processor);
+}
+function getPostProcessor(name) {
+  return postProcessors.get(name);
+}
+function listPostProcessors() {
+  return Array.from(postProcessors.keys());
+}
+async function unregisterPostProcessor(name) {
+  const processor = postProcessors.get(name);
+  if (!processor) {
+    const available = Array.from(postProcessors.keys());
+    const availableStr = available.length > 0 ? ` Available: ${available.join(", ")}` : "";
+    throw new Error(`Post-processor "${name}" is not registered.${availableStr}`);
+  }
+  try {
+    if (processor.shutdown) {
+      await processor.shutdown();
+    }
+  } catch (error) {
+    console.warn(`Error during shutdown of post-processor "${name}":`, error);
+  }
+  postProcessors.delete(name);
+}
+async function clearPostProcessors() {
+  const entries = Array.from(postProcessors.entries());
+  for (const [_name, processor] of entries) {
+    try {
+      if (processor.shutdown) {
+        await processor.shutdown();
+      }
+    } catch (error) {
+      console.warn(`Error during shutdown of post-processor "${_name}":`, error);
+    }
+  }
+  postProcessors.clear();
+}
+function validateValidator(validator) {
+  if (validator === null || validator === void 0) {
+    throw new Error("Validator cannot be null or undefined");
+  }
+  const obj = validator;
+  if (typeof obj.name !== "function") {
+    throw new Error("Validator must implement name() method");
+  }
+  if (typeof obj.validate !== "function") {
+    throw new Error("Validator must implement validate() method");
+  }
+  const name = obj.name();
+  if (typeof name !== "string" || name.trim() === "") {
+    throw new Error("Validator name must be a non-empty string");
+  }
+  return true;
+}
+function registerValidator(validator) {
+  validateValidator(validator);
+  const name = validator.name();
+  if (validators.has(name)) {
+    console.warn(`Validator "${name}" already registered, overwriting with new implementation`);
+  }
+  validators.set(name, validator);
+}
+function getValidator(name) {
+  return validators.get(name);
+}
+function listValidators() {
+  return Array.from(validators.keys());
+}
+async function unregisterValidator(name) {
+  const validator = validators.get(name);
+  if (!validator) {
+    const available = Array.from(validators.keys());
+    const availableStr = available.length > 0 ? ` Available: ${available.join(", ")}` : "";
+    throw new Error(`Validator "${name}" is not registered.${availableStr}`);
+  }
+  try {
+    if (validator.shutdown) {
+      await validator.shutdown();
+    }
+  } catch (error) {
+    console.warn(`Error during shutdown of validator "${name}":`, error);
+  }
+  validators.delete(name);
+}
+async function clearValidators() {
+  const entries = Array.from(validators.entries());
+  for (const [_name, validator] of entries) {
+    try {
+      if (validator.shutdown) {
+        await validator.shutdown();
+      }
+    } catch (error) {
+      console.warn(`Error during shutdown of validator "${_name}":`, error);
+    }
+  }
+  validators.clear();
+}
+function executePostProcessor(name, result) {
+  const processor = postProcessors.get(name);
+  if (!processor) {
+    return Promise.reject(new Error(`Post-processor "${name}" is not registered`));
+  }
+  try {
+    const output = processor.process(result);
+    if (output instanceof Promise) {
+      return output;
+    }
+    return Promise.resolve(output);
+  } catch (error) {
+    return Promise.reject(new Error(`Error executing post-processor "${name}": ${String(error)}`));
+  }
+}
+function executeValidator(name, result) {
+  const validator = validators.get(name);
+  if (!validator) {
+    return Promise.reject(new Error(`Validator "${name}" is not registered`));
+  }
+  try {
+    const output = validator.validate(result);
+    if (output instanceof Promise) {
+      return output;
+    }
+    return Promise.resolve(output);
+  } catch (error) {
+    return Promise.reject(new Error(`Error executing validator "${name}": ${String(error)}`));
+  }
+}
+function setupGlobalCallbacks() {
+  if (typeof globalThis !== "undefined") {
+    const callbacksObj = globalThis;
+    callbacksObj.__kreuzberg_execute_post_processor = executePostProcessor;
+    callbacksObj.__kreuzberg_execute_validator = executeValidator;
+  }
+}
+setupGlobalCallbacks();
+
 // typescript/index.ts
 var wasm = null;
 var initialized = false;
@@ -924,7 +1068,7 @@ async function initializePdfiumAsync(wasmModule) {
     return;
   }
   try {
-    const pdfiumModule = await import("./pdfium.js");
+    const pdfiumModule = await Promise.resolve().then(() => (init_pdfium(), pdfium_exports));
     const pdfium = typeof pdfiumModule.default === "function" ? await pdfiumModule.default() : pdfiumModule;
     const success = wasmModule.initialize_pdfium_render(pdfium, wasmModule, false);
     if (!success) {