@kreuzberg/wasm 4.0.0-rc.23 → 4.0.0-rc.25

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

@@ -1,5 +1,3 @@
-import { a as ExtractionConfig, E as ExtractionResult } from '../types-wVLLDHkl.js';
-
 /**
  * WASM Type Adapter
  *
@@ -27,7 +25,7 @@ import { a as ExtractionConfig, E as ExtractionResult } from '../types-wVLLDHkl.
  * const normalized = configToJS(config);
  * ```
  */
-
+import type { ExtractionConfig, ExtractionResult } from "../types.d.ts";
 /**
  * Convert a File or Blob to Uint8Array
  *
@@ -45,7 +43,7 @@ import { a as ExtractionConfig, E as ExtractionResult } from '../types-wVLLDHkl.
  * const result = await extractBytes(bytes, 'application/pdf');
  * ```
  */
-declare function fileToUint8Array(file: File | Blob): Promise<Uint8Array>;
+export declare function fileToUint8Array(file: File | Blob): Promise<Uint8Array>;
 /**
  * Normalize ExtractionConfig for WASM processing
  *
@@ -64,7 +62,7 @@ declare function fileToUint8Array(file: File | Blob): Promise<Uint8Array>;
  * const wasmConfig = configToJS(config);
  * ```
  */
-declare function configToJS(config: ExtractionConfig | null): Record<string, unknown>;
+export declare function configToJS(config: ExtractionConfig | null): Record<string, unknown>;
 /**
  * Parse WASM extraction result and convert to TypeScript type
  *
@@ -82,7 +80,7 @@ declare function configToJS(config: ExtractionConfig | null): Record<string, unk
  * console.log(result.content);
  * ```
  */
-declare function jsToExtractionResult(jsValue: unknown): ExtractionResult;
+export declare function jsToExtractionResult(jsValue: unknown): ExtractionResult;
 /**
  * Wrap and format WASM errors with context
  *
@@ -104,7 +102,7 @@ declare function jsToExtractionResult(jsValue: unknown): ExtractionResult;
  * }
  * ```
  */
-declare function wrapWasmError(error: unknown, context: string): Error;
+export declare function wrapWasmError(error: unknown, context: string): Error;
 /**
  * Validate that a WASM-returned value conforms to ExtractionResult structure
  *
@@ -116,6 +114,5 @@ declare function wrapWasmError(error: unknown, context: string): Error;
  *
  * @internal
  */
-declare function isValidExtractionResult(value: unknown): value is ExtractionResult;
-
-export { configToJS, fileToUint8Array, isValidExtractionResult, jsToExtractionResult, wrapWasmError };
+export declare function isValidExtractionResult(value: unknown): value is ExtractionResult;
+//# sourceMappingURL=wasm-adapter.d.ts.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"wasm-adapter.d.ts","sourceRoot":"","sources":["../../typescript/adapters/wasm-adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;GA0BG;AAEH,OAAO,KAAK,EAAyB,gBAAgB,EAAE,gBAAgB,EAAmB,MAAM,aAAa,CAAC;AAoC9G;;;;;;;;;;;;;;;;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,CAwMvE;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

@@ -104,30 +104,52 @@ function jsToExtractionResult(jsValue) {
       }
       embedding = c.embedding;
     }
-    if (typeof metadata.charStart !== "number") {
-      throw new Error("Invalid chunk metadata: charStart must be a number");
-    }
-    if (typeof metadata.charEnd !== "number") {
-      throw new Error("Invalid chunk metadata: charEnd must be a number");
-    }
-    if (!isNumberOrNull(metadata.tokenCount)) {
-      throw new Error("Invalid chunk metadata: tokenCount must be a number or null");
-    }
-    if (typeof metadata.chunkIndex !== "number") {
-      throw new Error("Invalid chunk metadata: chunkIndex must be a number");
-    }
-    if (typeof metadata.totalChunks !== "number") {
-      throw new Error("Invalid chunk metadata: totalChunks must be a number");
+    const coerceToNumber = (value, fieldName) => {
+      if (typeof value === "number") {
+        return value;
+      }
+      if (typeof value === "bigint") {
+        return Number(value);
+      }
+      if (typeof value === "string") {
+        const parsed = parseInt(value, 10);
+        if (isNaN(parsed)) {
+          throw new Error(`Invalid chunk metadata: ${fieldName} must be a valid number, got "${value}"`);
+        }
+        return parsed;
+      }
+      throw new Error(`Invalid chunk metadata: ${fieldName} must be a number, got ${typeof value}`);
+    };
+    const charStart = coerceToNumber(
+      metadata.charStart ?? metadata.char_start ?? metadata.byteStart ?? metadata.byte_start,
+      "charStart"
+    );
+    const charEnd = coerceToNumber(
+      metadata.charEnd ?? metadata.char_end ?? metadata.byteEnd ?? metadata.byte_end,
+      "charEnd"
+    );
+    const chunkIndex = coerceToNumber(
+      metadata.chunkIndex ?? metadata.chunk_index,
+      "chunkIndex"
+    );
+    const totalChunks = coerceToNumber(
+      metadata.totalChunks ?? metadata.total_chunks,
+      "totalChunks"
+    );
+    let tokenCount = null;
+    const tokenCountValue = metadata.tokenCount ?? metadata.token_count;
+    if (tokenCountValue !== null && tokenCountValue !== void 0) {
+      tokenCount = coerceToNumber(tokenCountValue, "tokenCount");
     }
     return {
       content: c.content,
       embedding,
       metadata: {
-        charStart: metadata.charStart,
-        charEnd: metadata.charEnd,
-        tokenCount: metadata.tokenCount,
-        chunkIndex: metadata.chunkIndex,
-        totalChunks: metadata.totalChunks
+        charStart,
+        charEnd,
+        tokenCount,
+        chunkIndex,
+        totalChunks
       }
     };
   }) : null;

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\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"]}
+{"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\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 (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(\n\t\t\t\t\tmetadata.chunkIndex ?? metadata.chunk_index,\n\t\t\t\t\t\"chunkIndex\"\n\t\t\t\t);\n\t\t\t\tconst totalChunks = coerceToNumber(\n\t\t\t\t\tmetadata.totalChunks ?? metadata.total_chunks,\n\t\t\t\t\t\"totalChunks\"\n\t\t\t\t);\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\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;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,MAAM,MAAM,GAAG;AAClB,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;AAAA,MAClB,SAAS,cAAc,SAAS;AAAA,MAChC;AAAA,IACD;AACA,UAAM,cAAc;AAAA,MACnB,SAAS,eAAe,SAAS;AAAA,MACjC;AAAA,IACD;AAEA,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;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.d.ts

@@ -1,226 +1,3 @@
-import { E as ExtractionConfig, a as ExtractionResult } from './types-wVLLDHkl.d.ts';
-export { C as Chunk, b as ChunkingConfig, c as ChunkMetadata, d as ExtractedImage, I as ImageExtractionConfig, L as LanguageDetectionConfig, M as Metadata, O as OcrBackendProtocol, e as OcrConfig, P as PageContent, f as PageExtractionConfig, g as PdfConfig, h as PostProcessorConfig, T as Table, i as TesseractConfig, j as TokenReductionConfig, E as ExtractionConfig, a as ExtractionResult } from './types-wVLLDHkl.d.ts';
-export { configToJS, fileToUint8Array, isValidExtractionResult, jsToExtractionResult, wrapWasmError } from './adapters/wasm-adapter.js';
-export { clearOcrBackends, getOcrBackend, listOcrBackends, registerOcrBackend, unregisterOcrBackend } from './ocr/registry.js';
-export { TesseractWasmBackend } from './ocr/tesseract-wasm-backend.js';
-export { type RuntimeType, type WasmCapabilities, detectRuntime, getRuntimeInfo, getRuntimeVersion, getWasmCapabilities, hasBigInt, hasBlob, hasFileApi, hasModuleWorkers, hasSharedArrayBuffer, hasWasm, hasWasmStreaming, hasWorkers, isBrowser, isBun, isDeno, isNode, isServerEnvironment, isWebEnvironment } from './runtime.d.ts';
-
-/**
- * Plugin Registry Module
- *
- * This module manages registrations and execution of post-processors and validators
- * for document extraction pipelines.
- *
- * # Thread Safety
- * All registrations are stored in Maps and are single-threaded safe for WASM environments.
- *
- * # Global Callback Functions
- * The WASM module can invoke processing via global callback functions:
- * - `__kreuzberg_execute_post_processor`: Execute a registered post-processor
- * - `__kreuzberg_execute_validator`: Execute a registered validator
- */
-
-/**
- * Post-processor plugin interface
- *
- * A post-processor modifies extraction results after extraction completes.
- */
-interface PostProcessor {
-    /**
-     * Get the processor name (must be non-empty string)
-     */
-    name(): string;
-    /**
-     * Get the processing stage (optional, defaults to "middle")
-     * - "early": Process early in the pipeline
-     * - "middle": Process in the middle of the pipeline
-     * - "late": Process late in the pipeline
-     */
-    stage?(): "early" | "middle" | "late";
-    /**
-     * Process an extraction result
-     * Can be sync or async
-     */
-    process(result: ExtractionResult): ExtractionResult | Promise<ExtractionResult>;
-    /**
-     * Shutdown the processor (optional)
-     */
-    shutdown?(): void | Promise<void>;
-}
-/**
- * Validator plugin interface
- *
- * A validator checks extraction results for correctness
- */
-interface Validator {
-    /**
-     * Get the validator name (must be non-empty string)
-     */
-    name(): string;
-    /**
-     * Get the validation priority (optional, defaults to 50)
-     * Higher numbers = higher priority (execute first)
-     */
-    priority?(): number;
-    /**
-     * Validate an extraction result
-     * Can be sync or async
-     */
-    validate(result: ExtractionResult): {
-        valid: boolean;
-        errors: string[];
-    } | Promise<{
-        valid: boolean;
-        errors: string[];
-    }>;
-    /**
-     * Shutdown the validator (optional)
-     */
-    shutdown?(): void | Promise<void>;
-}
-/**
- * Register a post-processor plugin
- *
- * @param processor - The post-processor to register
- * @throws {Error} If the processor is invalid or missing required methods
- *
- * @example
- * ```typescript
- * const processor = {
- *   name: () => "my-processor",
- *   stage: () => "middle",
- *   process: async (result) => {
- *     result.content = result.content.toUpperCase();
- *     return result;
- *   }
- * };
- * registerPostProcessor(processor);
- * ```
- */
-declare function registerPostProcessor(processor: PostProcessor): void;
-/**
- * Get a registered post-processor by name
- *
- * @param name - The processor name
- * @returns The processor, or undefined if not found
- *
- * @example
- * ```typescript
- * const processor = getPostProcessor("my-processor");
- * if (processor) {
- *   console.log("Found processor:", processor.name());
- * }
- * ```
- */
-declare function getPostProcessor(name: string): PostProcessor | undefined;
-/**
- * List all registered post-processor names
- *
- * @returns Array of processor names
- *
- * @example
- * ```typescript
- * const names = listPostProcessors();
- * console.log("Registered processors:", names);
- * ```
- */
-declare function listPostProcessors(): string[];
-/**
- * Unregister a post-processor and call its shutdown method
- *
- * @param name - The processor name
- * @throws {Error} If the processor is not registered
- *
- * @example
- * ```typescript
- * await unregisterPostProcessor("my-processor");
- * ```
- */
-declare function unregisterPostProcessor(name: string): Promise<void>;
-/**
- * Clear all registered post-processors
- *
- * Calls shutdown on all processors before clearing.
- *
- * @example
- * ```typescript
- * await clearPostProcessors();
- * ```
- */
-declare function clearPostProcessors(): Promise<void>;
-/**
- * Register a validator plugin
- *
- * @param validator - The validator to register
- * @throws {Error} If the validator is invalid or missing required methods
- *
- * @example
- * ```typescript
- * const validator = {
- *   name: () => "my-validator",
- *   priority: () => 50,
- *   validate: async (result) => {
- *     if (!result.content) {
- *       return { valid: false, errors: ["Content is empty"] };
- *     }
- *     return { valid: true, errors: [] };
- *   }
- * };
- * registerValidator(validator);
- * ```
- */
-declare function registerValidator(validator: Validator): void;
-/**
- * Get a registered validator by name
- *
- * @param name - The validator name
- * @returns The validator, or undefined if not found
- *
- * @example
- * ```typescript
- * const validator = getValidator("my-validator");
- * if (validator) {
- *   console.log("Found validator:", validator.name());
- * }
- * ```
- */
-declare function getValidator(name: string): Validator | undefined;
-/**
- * List all registered validator names
- *
- * @returns Array of validator names
- *
- * @example
- * ```typescript
- * const names = listValidators();
- * console.log("Registered validators:", names);
- * ```
- */
-declare function listValidators(): string[];
-/**
- * Unregister a validator and call its shutdown method
- *
- * @param name - The validator name
- * @throws {Error} If the validator is not registered
- *
- * @example
- * ```typescript
- * await unregisterValidator("my-validator");
- * ```
- */
-declare function unregisterValidator(name: string): Promise<void>;
-/**
- * Clear all registered validators
- *
- * Calls shutdown on all validators before clearing.
- *
- * @example
- * ```typescript
- * await clearValidators();
- * ```
- */
-declare function clearValidators(): Promise<void>;
-
 /**
  * Kreuzberg - WebAssembly Bindings for Browser and Runtime Environments
  *
@@ -316,8 +93,15 @@ declare function clearValidators(): Promise<void>;
  * const result = await extractBytes(bytes, 'application/pdf', config);
  * ```
  */
-
-declare function initWasm(): Promise<void>;
+import type { ExtractionConfig as ExtractionConfigType, ExtractionResult } from "./types.d.ts";
+export { configToJS, fileToUint8Array, isValidExtractionResult, jsToExtractionResult, wrapWasmError, } from "./adapters/wasm-adapter.d.ts";
+export { clearOcrBackends, getOcrBackend, listOcrBackends, registerOcrBackend, unregisterOcrBackend, } from "./ocr/registry.d.ts";
+export { TesseractWasmBackend } from "./ocr/tesseract-wasm-backend.d.ts";
+export { clearPostProcessors, clearValidators, getPostProcessor, getValidator, listPostProcessors, listValidators, type PostProcessor, registerPostProcessor, registerValidator, unregisterPostProcessor, unregisterValidator, type Validator, } from "./plugin-registry.d.ts";
+export { detectRuntime, getRuntimeInfo, getRuntimeVersion, getWasmCapabilities, hasBigInt, hasBlob, hasFileApi, hasModuleWorkers, hasSharedArrayBuffer, hasWasm, hasWasmStreaming, hasWorkers, isBrowser, isBun, isDeno, isNode, isServerEnvironment, isWebEnvironment, type RuntimeType, type WasmCapabilities, } from "./runtime.d.ts";
+export type * from "./types.d.ts";
+export type { Chunk, ChunkingConfig, ChunkMetadata, ExtractedImage, ExtractionConfig, ExtractionResult, ImageExtractionConfig, LanguageDetectionConfig, Metadata, OcrBackendProtocol, OcrConfig, PageContent, PageExtractionConfig, PdfConfig, PostProcessorConfig, Table, TesseractConfig, TokenReductionConfig, } from "./types.d.ts";
+export declare function initWasm(): Promise<void>;
 /**
  * Check if WASM module is initialized
  *
@@ -330,7 +114,7 @@ declare function initWasm(): Promise<void>;
  * }
  * ```
  */
-declare function isInitialized(): boolean;
+export declare function isInitialized(): boolean;
 /**
  * Get WASM module version
  *
@@ -343,7 +127,7 @@ declare function isInitialized(): boolean;
  * console.log(`Using Kreuzberg ${version}`);
  * ```
  */
-declare function getVersion(): string;
+export declare function getVersion(): string;
 /**
  * Get initialization error if module failed to load
  *
@@ -351,7 +135,7 @@ declare function getVersion(): string;
  *
  * @internal
  */
-declare function getInitializationError(): Error | null;
+export declare function getInitializationError(): Error | null;
 /**
  * Extract content from bytes (document data)
  *
@@ -393,7 +177,7 @@ declare function getInitializationError(): Error | null;
  * const result = await extractBytes(bytes, file.type);
  * ```
  */
-declare function extractBytes(data: Uint8Array, mimeType: string, config?: ExtractionConfig | null): Promise<ExtractionResult>;
+export declare function extractBytes(data: Uint8Array, mimeType: string, config?: ExtractionConfigType | null): Promise<ExtractionResult>;
 /**
  * Extract content from a file on the file system
  *
@@ -429,7 +213,7 @@ declare function extractBytes(data: Uint8Array, mimeType: string, config?: Extra
  * });
  * ```
  */
-declare function extractFile(path: string, mimeType?: string | null, config?: ExtractionConfig | null): Promise<ExtractionResult>;
+export declare function extractFile(path: string, mimeType?: string | null, config?: ExtractionConfigType | null): Promise<ExtractionResult>;
 /**
  * Extract content from a File or Blob (browser-friendly wrapper)
  *
@@ -462,7 +246,7 @@ declare function extractFile(path: string, mimeType?: string | null, config?: Ex
  * });
  * ```
  */
-declare function extractFromFile(file: File | Blob, mimeType?: string | null, config?: ExtractionConfig | null): Promise<ExtractionResult>;
+export declare function extractFromFile(file: File | Blob, mimeType?: string | null, config?: ExtractionConfigType | null): Promise<ExtractionResult>;
 /**
  * Extract content from bytes synchronously
  *
@@ -482,7 +266,7 @@ declare function extractFromFile(file: File | Blob, mimeType?: string | null, co
  * console.log(result.content);
  * ```
  */
-declare function extractBytesSync(data: Uint8Array, mimeType: string, config?: ExtractionConfig | null): ExtractionResult;
+export declare function extractBytesSync(data: Uint8Array, mimeType: string, config?: ExtractionConfigType | null): ExtractionResult;
 /**
  * Batch extract content from multiple byte arrays asynchronously
  *
@@ -504,10 +288,10 @@ declare function extractBytesSync(data: Uint8Array, mimeType: string, config?: E
  * results.forEach((result) => console.log(result.content));
  * ```
  */
-declare function batchExtractBytes(files: Array<{
+export declare function batchExtractBytes(files: Array<{
     data: Uint8Array;
     mimeType: string;
-}>, config?: ExtractionConfig | null): Promise<ExtractionResult[]>;
+}>, config?: ExtractionConfigType | null): Promise<ExtractionResult[]>;
 /**
  * Batch extract content from multiple byte arrays synchronously
  *
@@ -529,10 +313,10 @@ declare function batchExtractBytes(files: Array<{
  * results.forEach((result) => console.log(result.content));
  * ```
  */
-declare function batchExtractBytesSync(files: Array<{
+export declare function batchExtractBytesSync(files: Array<{
     data: Uint8Array;
     mimeType: string;
-}>, config?: ExtractionConfig | null): ExtractionResult[];
+}>, config?: ExtractionConfigType | null): ExtractionResult[];
 /**
  * Batch extract content from multiple File objects asynchronously
  *
@@ -554,7 +338,7 @@ declare function batchExtractBytesSync(files: Array<{
  * });
  * ```
  */
-declare function batchExtractFiles(files: File[], config?: ExtractionConfig | null): Promise<ExtractionResult[]>;
+export declare function batchExtractFiles(files: File[], config?: ExtractionConfigType | null): Promise<ExtractionResult[]>;
 /**
  * Enable OCR functionality with tesseract-wasm backend
  *
@@ -634,6 +418,5 @@ declare function batchExtractFiles(files: File[], config?: ExtractionConfig | nu
  * });
  * ```
  */
-declare function enableOcr(): Promise<void>;
-
-export { type PostProcessor, type Validator, batchExtractBytes, batchExtractBytesSync, batchExtractFiles, clearPostProcessors, clearValidators, enableOcr, extractBytes, extractBytesSync, extractFile, extractFromFile, getInitializationError, getPostProcessor, getValidator, getVersion, initWasm, isInitialized, listPostProcessors, listValidators, registerPostProcessor, registerValidator, unregisterPostProcessor, unregisterValidator };
+export declare function enableOcr(): Promise<void>;
+//# sourceMappingURL=index.d.ts.map

package/dist/index.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../typescript/index.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA8FG;AAMH,OAAO,KAAK,EAAE,gBAAgB,IAAI,oBAAoB,EAAE,gBAAgB,EAAE,MAAM,YAAY,CAAC;AAE7F,OAAO,EACN,UAAU,EACV,gBAAgB,EAChB,uBAAuB,EACvB,oBAAoB,EACpB,aAAa,GACb,MAAM,4BAA4B,CAAC;AACpC,OAAO,EACN,gBAAgB,EAChB,aAAa,EACb,eAAe,EACf,kBAAkB,EAClB,oBAAoB,GACpB,MAAM,mBAAmB,CAAC;AAC3B,OAAO,EAAE,oBAAoB,EAAE,MAAM,iCAAiC,CAAC;AACvE,OAAO,EACN,mBAAmB,EACnB,eAAe,EACf,gBAAgB,EAChB,YAAY,EACZ,kBAAkB,EAClB,cAAc,EACd,KAAK,aAAa,EAClB,qBAAqB,EACrB,iBAAiB,EACjB,uBAAuB,EACvB,mBAAmB,EACnB,KAAK,SAAS,GACd,MAAM,sBAAsB,CAAC;AAC9B,OAAO,EACN,aAAa,EACb,cAAc,EACd,iBAAiB,EACjB,mBAAmB,EACnB,SAAS,EACT,OAAO,EACP,UAAU,EACV,gBAAgB,EAChB,oBAAoB,EACpB,OAAO,EACP,gBAAgB,EAChB,UAAU,EACV,SAAS,EACT,KAAK,EACL,MAAM,EACN,MAAM,EACN,mBAAmB,EACnB,gBAAgB,EAChB,KAAK,WAAW,EAChB,KAAK,gBAAgB,GACrB,MAAM,cAAc,CAAC;AACtB,mBAAmB,YAAY,CAAC;AAChC,YAAY,EACX,KAAK,EACL,cAAc,EACd,aAAa,EACb,cAAc,EACd,gBAAgB,EAChB,gBAAgB,EAChB,qBAAqB,EACrB,uBAAuB,EACvB,QAAQ,EACR,kBAAkB,EAClB,SAAS,EACT,WAAW,EACX,oBAAoB,EACpB,SAAS,EACT,mBAAmB,EACnB,KAAK,EACL,eAAe,EACf,oBAAoB,GACpB,MAAM,YAAY,CAAC;AA4IpB,wBAAsB,QAAQ,IAAI,OAAO,CAAC,IAAI,CAAC,CA6C9C;AAED;;;;;;;;;;;GAWG;AACH,wBAAgB,aAAa,IAAI,OAAO,CAEvC;AAED;;;;;;;;;;;GAWG;AACH,wBAAgB,UAAU,IAAI,MAAM,CAUnC;AAED;;;;;;GAMG;AACH,wBAAgB,sBAAsB,IAAI,KAAK,GAAG,IAAI,CAErD;AAED;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAwCG;AACH,wBAAsB,YAAY,CACjC,IAAI,EAAE,UAAU,EAChB,QAAQ,EAAE,MAAM,EAChB,MAAM,CAAC,EAAE,oBAAoB,GAAG,IAAI,GAClC,OAAO,CAAC,gBAAgB,CAAC,CA8B3B;AAED;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAkCG;AACH,wBAAsB,WAAW,CAChC,IAAI,EAAE,MAAM,EACZ,QAAQ,CAAC,EAAE,MAAM,GAAG,IAAI,EACxB,MAAM,CAAC,EAAE,oBAAoB,GAAG,IAAI,GAClC,OAAO,CAAC,gBAAgB,CAAC,CAqD3B;AAED;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA+BG;AACH,wBAAsB,eAAe,CACpC,IAAI,EAAE,IAAI,GAAG,IAAI,EACjB,QAAQ,CAAC,EAAE,MAAM,GAAG,IAAI,EACxB,MAAM,CAAC,EAAE,oBAAoB,GAAG,IAAI,GAClC,OAAO,CAAC,gBAAgB,CAAC,CAmB3B;AAED;;;;;;;;;;;;;;;;;;GAkBG;AACH,wBAAgB,gBAAgB,CAC/B,IAAI,EAAE,UAAU,EAChB,QAAQ,EAAE,MAAM,EAChB,MAAM,CAAC,EAAE,oBAAoB,GAAG,IAAI,GAClC,gBAAgB,CA8BlB;AAED;;;;;;;;;;;;;;;;;;;;GAoBG;AACH,wBAAsB,iBAAiB,CACtC,KAAK,EAAE,KAAK,CAAC;IAAE,IAAI,EAAE,UAAU,CAAC;IAAC,QAAQ,EAAE,MAAM,CAAA;CAAE,CAAC,EACpD,MAAM,CAAC,EAAE,oBAAoB,GAAG,IAAI,GAClC,OAAO,CAAC,gBAAgB,EAAE,CAAC,CA+D7B;AAED;;;;;;;;;;;;;;;;;;;;GAoBG;AACH,wBAAgB,qBAAqB,CACpC,KAAK,EAAE,KAAK,CAAC;IAAE,IAAI,EAAE,UAAU,CAAC;IAAC,QAAQ,EAAE,MAAM,CAAA;CAAE,CAAC,EACpD,MAAM,CAAC,EAAE,oBAAoB,GAAG,IAAI,GAClC,gBAAgB,EAAE,CA+DpB;AAED;;;;;;;;;;;;;;;;;;;;GAoBG;AACH,wBAAsB,iBAAiB,CACtC,KAAK,EAAE,IAAI,EAAE,EACb,MAAM,CAAC,EAAE,oBAAoB,GAAG,IAAI,GAClC,OAAO,CAAC,gBAAgB,EAAE,CAAC,CAiC7B;AAED;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA8EG;AACH,wBAAsB,SAAS,IAAI,OAAO,CAAC,IAAI,CAAC,CAoB/C"}

package/dist/index.js

@@ -130,30 +130,52 @@ function jsToExtractionResult(jsValue) {
       }
       embedding = c.embedding;
     }
-    if (typeof metadata.charStart !== "number") {
-      throw new Error("Invalid chunk metadata: charStart must be a number");
-    }
-    if (typeof metadata.charEnd !== "number") {
-      throw new Error("Invalid chunk metadata: charEnd must be a number");
-    }
-    if (!isNumberOrNull(metadata.tokenCount)) {
-      throw new Error("Invalid chunk metadata: tokenCount must be a number or null");
-    }
-    if (typeof metadata.chunkIndex !== "number") {
-      throw new Error("Invalid chunk metadata: chunkIndex must be a number");
-    }
-    if (typeof metadata.totalChunks !== "number") {
-      throw new Error("Invalid chunk metadata: totalChunks must be a number");
+    const coerceToNumber = (value, fieldName) => {
+      if (typeof value === "number") {
+        return value;
+      }
+      if (typeof value === "bigint") {
+        return Number(value);
+      }
+      if (typeof value === "string") {
+        const parsed = parseInt(value, 10);
+        if (isNaN(parsed)) {
+          throw new Error(`Invalid chunk metadata: ${fieldName} must be a valid number, got "${value}"`);
+        }
+        return parsed;
+      }
+      throw new Error(`Invalid chunk metadata: ${fieldName} must be a number, got ${typeof value}`);
+    };
+    const charStart = coerceToNumber(
+      metadata.charStart ?? metadata.char_start ?? metadata.byteStart ?? metadata.byte_start,
+      "charStart"
+    );
+    const charEnd = coerceToNumber(
+      metadata.charEnd ?? metadata.char_end ?? metadata.byteEnd ?? metadata.byte_end,
+      "charEnd"
+    );
+    const chunkIndex = coerceToNumber(
+      metadata.chunkIndex ?? metadata.chunk_index,
+      "chunkIndex"
+    );
+    const totalChunks = coerceToNumber(
+      metadata.totalChunks ?? metadata.total_chunks,
+      "totalChunks"
+    );
+    let tokenCount = null;
+    const tokenCountValue = metadata.tokenCount ?? metadata.token_count;
+    if (tokenCountValue !== null && tokenCountValue !== void 0) {
+      tokenCount = coerceToNumber(tokenCountValue, "tokenCount");
     }
     return {
       content: c.content,
       embedding,
       metadata: {
-        charStart: metadata.charStart,
-        charEnd: metadata.charEnd,
-        tokenCount: metadata.tokenCount,
-        chunkIndex: metadata.chunkIndex,
-        totalChunks: metadata.totalChunks
+        charStart,
+        charEnd,
+        tokenCount,
+        chunkIndex,
+        totalChunks
       }
     };
   }) : null;