@xberg-io/xberg-wasm 1.3.6 → 1.3.9

package/package.json

@@ -1,6 +1,16 @@
 {
+  "author": {
+    "name": "Na'aman Hirschfeld",
+    "email": "naaman@xberg.io"
+  },
+  "contributors": [
+    {
+      "name": "Tobias Perlstein",
+      "email": "tobias@xberg.io"
+    }
+  ],
   "name": "@xberg-io/xberg-wasm",
-  "version": "1.3.6",
+  "version": "1.3.9",
   "private": false,
   "description": "High-performance document intelligence library",
   "license": "MIT",

package/pkg/web/package.json

@@ -1,8 +1,12 @@
 {
   "name": "xberg-wasm",
   "type": "module",
+  "collaborators": [
+    "Na'aman Hirschfeld <naaman@xberg.io>",
+    "Tobias Perlstein <tobias@xberg.io>"
+  ],
   "description": "High-performance document intelligence library",
-  "version": "1.3.6",
+  "version": "1.3.9",
   "license": "MIT",
   "repository": {
     "type": "git",

package/pkg/web/xberg_wasm.d.ts

@@ -3337,11 +3337,11 @@ export class WasmExtractedDocument {
      * of those handles is dead once this returns. Passing a freshly built value is fine. To
      * keep a handle you already hold, build with `default()` and assign the property instead:
      * the generated setter borrows and leaves your handle alive.
-     * Takes ownership of every element in `tables`, `detectedLanguageConfidences`, `chunks`, `images`, `pages`, `elements`, `ocrElements`, `extractedKeywords`, `processingWarnings`, `annotations`, `children`, `uris`, `revisions`, `llmUsage`, `entities`, `pageClassifications`, `formulas`, `formFields`.
+     * Takes ownership of every element in `tables`, `detectedLanguageConfidences`, `chunks`, `images`, `pages`, `elements`, `ocrElements`, `extractedKeywords`, `processingWarnings`, `annotations`, `children`, `uris`, `revisions`, `llmUsage`, `entities`, `pageClassifications`, `formulas`, `formFields`, `ocrPageFailures`.
      * wasm-bindgen cannot borrow class values nested in an array. To keep those handles,
      * copy each retained element with `copyForTransfer()` before passing the array.
      */
-    constructor(content?: string | null, mimeType?: string | null, metadata?: WasmMetadata | null, tables?: WasmTable[] | null, counts?: WasmDocumentCounts | null, processingWarnings?: WasmProcessingWarning[] | null, formulas?: WasmFormula[] | null, formFields?: WasmPdfFormField[] | null, extractionMethod?: WasmExtractionMethod | null, detectedLanguages?: string[] | null, detectedLanguageConfidences?: WasmLanguageConfidence[] | null, chunks?: WasmChunk[] | null, images?: WasmExtractedImage[] | null, pages?: WasmPageContent[] | null, elements?: WasmElement[] | null, djotContent?: WasmDjotContent | null, ocrElements?: WasmOcrElement[] | null, document?: WasmDocumentStructure | null, extractedKeywords?: WasmKeyword[] | null, qualityScore?: number | null, annotations?: WasmPdfAnnotation[] | null, children?: WasmArchiveEntry[] | null, uris?: WasmExtractedUri[] | null, revisions?: WasmDocumentRevision[] | null, structuredOutput?: any | null, llmUsage?: WasmLlmUsage[] | null, entities?: WasmEntity[] | null, summary?: WasmDocumentSummary | null, extractionConfidence?: WasmExtractionConfidence | null, translation?: WasmTranslation | null, pageClassifications?: WasmPageClassification[] | null, redactionReport?: WasmRedactionReport | null);
+    constructor(content?: string | null, mimeType?: string | null, metadata?: WasmMetadata | null, tables?: WasmTable[] | null, counts?: WasmDocumentCounts | null, processingWarnings?: WasmProcessingWarning[] | null, formulas?: WasmFormula[] | null, formFields?: WasmPdfFormField[] | null, extractionMethod?: WasmExtractionMethod | null, detectedLanguages?: string[] | null, detectedLanguageConfidences?: WasmLanguageConfidence[] | null, chunks?: WasmChunk[] | null, images?: WasmExtractedImage[] | null, pages?: WasmPageContent[] | null, elements?: WasmElement[] | null, djotContent?: WasmDjotContent | null, ocrElements?: WasmOcrElement[] | null, document?: WasmDocumentStructure | null, extractedKeywords?: WasmKeyword[] | null, qualityScore?: number | null, annotations?: WasmPdfAnnotation[] | null, children?: WasmArchiveEntry[] | null, uris?: WasmExtractedUri[] | null, revisions?: WasmDocumentRevision[] | null, structuredOutput?: any | null, llmUsage?: WasmLlmUsage[] | null, entities?: WasmEntity[] | null, summary?: WasmDocumentSummary | null, extractionConfidence?: WasmExtractionConfidence | null, translation?: WasmTranslation | null, pageClassifications?: WasmPageClassification[] | null, redactionReport?: WasmRedactionReport | null, ocrPageFailures?: WasmOcrPageFailure[] | null);
     /**
      * Returns detached copies of the elements in `annotations`.
      *
@@ -3530,6 +3530,19 @@ export class WasmExtractedDocument {
      * element with `copyForTransfer()` before passing the array.
      */
     set ocrElements(value: WasmOcrElement[] | null | undefined);
+    /**
+     * Returns detached copies of the elements in `ocrPageFailures`.
+     *
+     * Changes to the returned values do not update this object. Modify the copies,
+     * then assign the array back to `ocrPageFailures`.
+     */
+    get ocrPageFailures(): Array<any> | undefined;
+    /**
+     * Takes ownership of every element in `ocrPageFailures`.
+     * wasm-bindgen cannot borrow class values nested in an array; copy each retained
+     * element with `copyForTransfer()` before passing the array.
+     */
+    set ocrPageFailures(value: WasmOcrPageFailure[] | null | undefined);
     /**
      * Returns detached copies of the elements in `pageClassifications`.
      *
@@ -3895,7 +3908,7 @@ export class WasmExtractionConfig {
      * keep a handle you already hold, build with `default()` and assign the property instead:
      * the generated setter borrows and leaves your handle alive.
      */
-    constructor(mimeDetectionPolicy?: WasmMimeDetectionPolicy | null, useCache?: boolean | null, enableQualityProcessing?: boolean | null, forceOcr?: boolean | null, ocrStrategy?: any | null, disableOcr?: boolean | null, resultFormat?: WasmResultFormat | null, outputFormat?: any | null, escapeMarkdown?: boolean | null, tableAnchors?: boolean | null, jupyterCellRendering?: WasmJupyterCellRendering | null, applyNotebookCellTags?: boolean | null, useLayoutForMarkdown?: boolean | null, includeDocumentStructure?: boolean | null, url?: WasmUrlExtractionConfig | null, maxArchiveDepth?: number | null, ocr?: WasmOcrConfig | null, forceOcrPages?: Uint32Array | null, ocrNearEmptyFallback?: boolean | null, ocrScannedPageQualityGate?: boolean | null, ocrEmbeddedImages?: boolean | null, chunking?: WasmChunkingConfig | null, contentFilter?: WasmContentFilterConfig | null, images?: WasmImageExtractionConfig | null, pdfOptions?: WasmPdfConfig | null, tokenReduction?: WasmTokenReductionOptions | null, languageDetection?: WasmLanguageDetectionConfig | null, pages?: WasmPageConfig | null, keywords?: WasmKeywordConfig | null, postprocessor?: WasmPostProcessorConfig | null, htmlOptions?: WasmConversionOptions | null, htmlOutput?: WasmHtmlOutputConfig | null, extractionTimeoutSecs?: bigint | null, maxConcurrentExtractions?: number | null, securityLimits?: WasmSecurityLimits | null, maxEmbeddedFileBytes?: bigint | null, layout?: WasmLayoutDetectionConfig | null, transcription?: WasmTranscriptionConfig | null, acceleration?: WasmAccelerationConfig | null, cacheNamespace?: string | null, cacheTtlSecs?: bigint | null, email?: WasmEmailConfig | null, csv?: WasmCsvConfig | null, geojson?: WasmGeoJsonExtractionConfig | null, concurrency?: WasmConcurrencyConfig | null, structuredExtraction?: WasmStructuredExtractionConfig | null, ner?: WasmNerConfig | null, redaction?: WasmRedactionConfig | null, summarization?: WasmSummarizationConfig | null, translation?: WasmTranslationConfig | null, pageClassification?: WasmPageClassificationConfig | null, chunkClassification?: WasmChunkClassificationConfig | null, captioning?: WasmCaptioningConfig | null, qrCodes?: boolean | null);
+    constructor(mimeDetectionPolicy?: WasmMimeDetectionPolicy | null, useCache?: boolean | null, enableQualityProcessing?: boolean | null, forceOcr?: boolean | null, ocrStrategy?: any | null, disableOcr?: boolean | null, resultFormat?: WasmResultFormat | null, outputFormat?: any | null, escapeMarkdown?: boolean | null, tableAnchors?: boolean | null, jupyterCellRendering?: WasmJupyterCellRendering | null, applyNotebookCellTags?: boolean | null, useLayoutForMarkdown?: boolean | null, includeDocumentStructure?: boolean | null, strictParse?: boolean | null, url?: WasmUrlExtractionConfig | null, maxArchiveDepth?: number | null, ocr?: WasmOcrConfig | null, forceOcrPages?: Uint32Array | null, ocrNearEmptyFallback?: boolean | null, ocrScannedPageQualityGate?: boolean | null, ocrEmbeddedImages?: boolean | null, chunking?: WasmChunkingConfig | null, contentFilter?: WasmContentFilterConfig | null, images?: WasmImageExtractionConfig | null, pdfOptions?: WasmPdfConfig | null, tokenReduction?: WasmTokenReductionOptions | null, languageDetection?: WasmLanguageDetectionConfig | null, pages?: WasmPageConfig | null, keywords?: WasmKeywordConfig | null, postprocessor?: WasmPostProcessorConfig | null, htmlOptions?: WasmConversionOptions | null, htmlOutput?: WasmHtmlOutputConfig | null, extractionTimeoutSecs?: bigint | null, maxConcurrentExtractions?: number | null, securityLimits?: WasmSecurityLimits | null, maxEmbeddedFileBytes?: bigint | null, layout?: WasmLayoutDetectionConfig | null, transcription?: WasmTranscriptionConfig | null, acceleration?: WasmAccelerationConfig | null, cacheNamespace?: string | null, cacheTtlSecs?: bigint | null, email?: WasmEmailConfig | null, csv?: WasmCsvConfig | null, geojson?: WasmGeoJsonExtractionConfig | null, concurrency?: WasmConcurrencyConfig | null, structuredExtraction?: WasmStructuredExtractionConfig | null, ner?: WasmNerConfig | null, redaction?: WasmRedactionConfig | null, summarization?: WasmSummarizationConfig | null, translation?: WasmTranslationConfig | null, pageClassification?: WasmPageClassificationConfig | null, chunkClassification?: WasmChunkClassificationConfig | null, captioning?: WasmCaptioningConfig | null, qrCodes?: boolean | null);
     /**
      * Whether embedded images get OCR'd.
      *
@@ -4182,6 +4195,7 @@ export class WasmExtractionConfig {
      */
     get securityLimits(): WasmSecurityLimits | undefined;
     set securityLimits(value: WasmSecurityLimits);
+    strictParse: boolean;
     /**
      * Returns a detached copy of `structuredExtraction`.
      *
@@ -4423,7 +4437,7 @@ export class WasmFileExtractionConfig {
      * keep a handle you already hold, build with `default()` and assign the property instead:
      * the generated setter borrows and leaves your handle alive.
      */
-    constructor(mimeDetectionPolicy?: WasmMimeDetectionPolicy | null, enableQualityProcessing?: boolean | null, ocr?: WasmOcrConfig | null, forceOcr?: boolean | null, ocrStrategy?: any | null, forceOcrPages?: Uint32Array | null, disableOcr?: boolean | null, chunking?: WasmChunkingConfig | null, contentFilter?: WasmContentFilterConfig | null, images?: WasmImageExtractionConfig | null, pdfOptions?: WasmPdfConfig | null, tokenReduction?: WasmTokenReductionOptions | null, languageDetection?: WasmLanguageDetectionConfig | null, pages?: WasmPageConfig | null, keywords?: WasmKeywordConfig | null, postprocessor?: WasmPostProcessorConfig | null, htmlOptions?: WasmConversionOptions | null, htmlOutput?: WasmHtmlOutputConfig | null, resultFormat?: WasmResultFormat | null, outputFormat?: any | null, includeDocumentStructure?: boolean | null, layout?: WasmLayoutDetectionConfig | null, transcription?: WasmTranscriptionConfig | null, timeoutSecs?: bigint | null, structuredExtraction?: WasmStructuredExtractionConfig | null, url?: WasmUrlExtractionConfig | null, ner?: WasmNerConfig | null, redaction?: WasmRedactionConfig | null, summarization?: WasmSummarizationConfig | null, translation?: WasmTranslationConfig | null, pageClassification?: WasmPageClassificationConfig | null, chunkClassification?: WasmChunkClassificationConfig | null, captioning?: WasmCaptioningConfig | null, qrCodes?: boolean | null);
+    constructor(mimeDetectionPolicy?: WasmMimeDetectionPolicy | null, enableQualityProcessing?: boolean | null, ocr?: WasmOcrConfig | null, forceOcr?: boolean | null, ocrStrategy?: any | null, forceOcrPages?: Uint32Array | null, disableOcr?: boolean | null, chunking?: WasmChunkingConfig | null, contentFilter?: WasmContentFilterConfig | null, images?: WasmImageExtractionConfig | null, pdfOptions?: WasmPdfConfig | null, tokenReduction?: WasmTokenReductionOptions | null, languageDetection?: WasmLanguageDetectionConfig | null, pages?: WasmPageConfig | null, keywords?: WasmKeywordConfig | null, postprocessor?: WasmPostProcessorConfig | null, htmlOptions?: WasmConversionOptions | null, htmlOutput?: WasmHtmlOutputConfig | null, resultFormat?: WasmResultFormat | null, outputFormat?: any | null, includeDocumentStructure?: boolean | null, strictParse?: boolean | null, layout?: WasmLayoutDetectionConfig | null, transcription?: WasmTranscriptionConfig | null, timeoutSecs?: bigint | null, structuredExtraction?: WasmStructuredExtractionConfig | null, url?: WasmUrlExtractionConfig | null, ner?: WasmNerConfig | null, redaction?: WasmRedactionConfig | null, summarization?: WasmSummarizationConfig | null, translation?: WasmTranslationConfig | null, pageClassification?: WasmPageClassificationConfig | null, chunkClassification?: WasmChunkClassificationConfig | null, captioning?: WasmCaptioningConfig | null, qrCodes?: boolean | null);
     /**
      * Returns a detached copy of `captioning`.
      *
@@ -4580,6 +4594,8 @@ export class WasmFileExtractionConfig {
     set redaction(value: WasmRedactionConfig);
     get resultFormat(): string | undefined;
     set resultFormat(value: WasmResultFormat | null | undefined);
+    get strictParse(): boolean | undefined;
+    set strictParse(value: boolean | null | undefined);
     /**
      * Returns a detached copy of `structuredExtraction`.
      *
@@ -7343,6 +7359,32 @@ export class WasmOcrMetadata {
     set tableRows(value: number | null | undefined);
 }
 
+/**
+ * A page whose OCR failed while extraction of the document continued.
+ *
+ * The warning in `ExtractedDocument.processing_warnings` stays available. One warning
+ * can cover several pages: when an automatic OCR run of selected pages fails as a whole,
+ * it adds one warning and one record for each selected page. When the OCR fallback for the
+ * whole document fails, it adds one warning and one record for each page that has no native
+ * text. Repeated failures on one page make one record.
+ */
+export class WasmOcrPageFailure {
+    free(): void;
+    [Symbol.dispose](): void;
+    /**
+     * Return an independent handle for an ownership-taking API.
+     *
+     * Use `copyForTransfer()` for each class value placed in an array when the original
+     * handle must remain usable.
+     */
+    copyForTransfer(): WasmOcrPageFailure;
+    static default(): WasmOcrPageFailure;
+    constructor(page?: number | null, error?: string | null, recovered?: boolean | null);
+    error: string;
+    page: number;
+    recovered: boolean;
+}
+
 /**
  * Multi-backend OCR pipeline with quality-based fallback.
  *
@@ -7921,9 +7963,10 @@ export class WasmPageConfig {
      */
     copyForTransfer(): WasmPageConfig;
     static default(): WasmPageConfig;
-    constructor(extractPages?: boolean | null, insertPageMarkers?: boolean | null, markerFormat?: string | null);
+    constructor(extractPages?: boolean | null, insertPageMarkers?: boolean | null, markerFormat?: string | null, keepNativeContent?: boolean | null);
     extractPages: boolean;
     insertPageMarkers: boolean;
+    keepNativeContent: boolean;
     markerFormat: string;
 }
 
@@ -7967,7 +8010,7 @@ export class WasmPageContent {
      * wasm-bindgen cannot borrow class values nested in an array. To keep those handles,
      * copy each retained element with `copyForTransfer()` before passing the array.
      */
-    constructor(pageNumber: number, content: string, tables: WasmTable[], imageIndices: Uint32Array, imagePreprocessing?: WasmImagePreprocessingMetadata | null, hierarchy?: WasmPageHierarchy | null, isBlank?: boolean | null, layoutRegions?: WasmLayoutRegion[] | null, speakerNotes?: string | null, sectionName?: string | null, sheetName?: string | null, ocrConfidence?: WasmPageOcrConfidence | null);
+    constructor(pageNumber: number, content: string, tables: WasmTable[], imageIndices: Uint32Array, imagePreprocessing?: WasmImagePreprocessingMetadata | null, hierarchy?: WasmPageHierarchy | null, isBlank?: boolean | null, layoutRegions?: WasmLayoutRegion[] | null, speakerNotes?: string | null, sectionName?: string | null, sheetName?: string | null, ocrConfidence?: WasmPageOcrConfidence | null, nativeContent?: string | null);
     content: string;
     /**
      * Returns a detached copy of `hierarchy`.
@@ -8001,6 +8044,8 @@ export class WasmPageContent {
      * element with `copyForTransfer()` before passing the array.
      */
     set layoutRegions(value: WasmLayoutRegion[] | null | undefined);
+    get nativeContent(): string | undefined;
+    set nativeContent(value: string | null | undefined);
     /**
      * Returns a detached copy of `ocrConfidence`.
      *
@@ -11689,6 +11734,7 @@ export interface InitOutput {
     readonly __wbg_wasmocrelementconfig_free: (a: number, b: number) => void;
     readonly __wbg_wasmocrextractionresult_free: (a: number, b: number) => void;
     readonly __wbg_wasmocrmetadata_free: (a: number, b: number) => void;
+    readonly __wbg_wasmocrpagefailure_free: (a: number, b: number) => void;
     readonly __wbg_wasmocrpipelineconfig_free: (a: number, b: number) => void;
     readonly __wbg_wasmocrpipelinestage_free: (a: number, b: number) => void;
     readonly __wbg_wasmocrpoint_free: (a: number, b: number) => void;
@@ -13042,8 +13088,9 @@ export interface InitOutput {
     readonly wasmextracteddocument_llmUsage: (a: number) => any;
     readonly wasmextracteddocument_metadata: (a: number) => number;
     readonly wasmextracteddocument_mimeType: (a: number) => [number, number];
-    readonly wasmextracteddocument_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number, k: number, l: number, m: number, n: number, o: number, p: number, q: number, r: number, s: number, t: number, u: number, v: number, w: number, x: number, y: number, z: number, a1: number, b1: number, c1: number, d1: number, e1: number, f1: number, g1: number, h1: number, i1: number, j1: number, k1: number, l1: number, m1: number, n1: number, o1: number, p1: number, q1: number, r1: number, s1: number, t1: number, u1: number, v1: number, w1: number, x1: number, y1: number, z1: number, a2: number, b2: number) => number;
+    readonly wasmextracteddocument_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number, k: number, l: number, m: number, n: number, o: number, p: number, q: number, r: number, s: number, t: number, u: number, v: number, w: number, x: number, y: number, z: number, a1: number, b1: number, c1: number, d1: number, e1: number, f1: number, g1: number, h1: number, i1: number, j1: number, k1: number, l1: number, m1: number, n1: number, o1: number, p1: number, q1: number, r1: number, s1: number, t1: number, u1: number, v1: number, w1: number, x1: number, y1: number, z1: number, a2: number, b2: number, c2: number, d2: number) => number;
     readonly wasmextracteddocument_ocrElements: (a: number) => any;
+    readonly wasmextracteddocument_ocrPageFailures: (a: number) => any;
     readonly wasmextracteddocument_pageClassifications: (a: number) => any;
     readonly wasmextracteddocument_pages: (a: number) => any;
     readonly wasmextracteddocument_processingWarnings: (a: number) => [number, number];
@@ -13071,6 +13118,7 @@ export interface InitOutput {
     readonly wasmextracteddocument_set_metadata: (a: number, b: number) => void;
     readonly wasmextracteddocument_set_mimeType: (a: number, b: number, c: number) => void;
     readonly wasmextracteddocument_set_ocrElements: (a: number, b: number, c: number) => void;
+    readonly wasmextracteddocument_set_ocrPageFailures: (a: number, b: number, c: number) => void;
     readonly wasmextracteddocument_set_pageClassifications: (a: number, b: number, c: number) => void;
     readonly wasmextracteddocument_set_pages: (a: number, b: number, c: number) => void;
     readonly wasmextracteddocument_set_processingWarnings: (a: number, b: number, c: number) => void;
@@ -13236,7 +13284,7 @@ export interface InitOutput {
     readonly wasmextractionconfig_needsImageData: (a: number) => number;
     readonly wasmextractionconfig_needsImageProcessing: (a: number) => number;
     readonly wasmextractionconfig_ner: (a: number) => number;
-    readonly wasmextractionconfig_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number, k: number, l: number, m: number, n: number, o: number, p: number, q: number, r: number, s: number, t: number, u: number, v: number, w: number, x: number, y: number, z: number, a1: number, b1: number, c1: number, d1: number, e1: number, f1: number, g1: number, h1: number, i1: bigint, j1: number, k1: number, l1: number, m1: bigint, n1: number, o1: number, p1: number, q1: number, r1: number, s1: number, t1: bigint, u1: number, v1: number, w1: number, x1: number, y1: number, z1: number, a2: number, b2: number, c2: number, d2: number, e2: number, f2: number, g2: number) => number;
+    readonly wasmextractionconfig_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number, k: number, l: number, m: number, n: number, o: number, p: number, q: number, r: number, s: number, t: number, u: number, v: number, w: number, x: number, y: number, z: number, a1: number, b1: number, c1: number, d1: number, e1: number, f1: number, g1: number, h1: number, i1: number, j1: bigint, k1: number, l1: number, m1: number, n1: bigint, o1: number, p1: number, q1: number, r1: number, s1: number, t1: number, u1: bigint, v1: number, w1: number, x1: number, y1: number, z1: number, a2: number, b2: number, c2: number, d2: number, e2: number, f2: number, g2: number, h2: number) => number;
     readonly wasmextractionconfig_ocr: (a: number) => number;
     readonly wasmextractionconfig_ocrEmbeddedImages: (a: number) => number;
     readonly wasmextractionconfig_ocrNearEmptyFallback: (a: number) => number;
@@ -13297,6 +13345,7 @@ export interface InitOutput {
     readonly wasmextractionconfig_set_redaction: (a: number, b: number) => void;
     readonly wasmextractionconfig_set_resultFormat: (a: number, b: number) => void;
     readonly wasmextractionconfig_set_securityLimits: (a: number, b: number) => void;
+    readonly wasmextractionconfig_set_strictParse: (a: number, b: number) => void;
     readonly wasmextractionconfig_set_structuredExtraction: (a: number, b: number) => void;
     readonly wasmextractionconfig_set_summarization: (a: number, b: number) => void;
     readonly wasmextractionconfig_set_tableAnchors: (a: number, b: number) => void;
@@ -13306,6 +13355,7 @@ export interface InitOutput {
     readonly wasmextractionconfig_set_url: (a: number, b: number) => void;
     readonly wasmextractionconfig_set_useCache: (a: number, b: number) => void;
     readonly wasmextractionconfig_set_useLayoutForMarkdown: (a: number, b: number) => void;
+    readonly wasmextractionconfig_strictParse: (a: number) => number;
     readonly wasmextractionconfig_structuredExtraction: (a: number) => number;
     readonly wasmextractionconfig_summarization: (a: number) => number;
     readonly wasmextractionconfig_tableAnchors: (a: number) => number;
@@ -13412,7 +13462,7 @@ export interface InitOutput {
     readonly wasmfileextractionconfig_layout: (a: number) => number;
     readonly wasmfileextractionconfig_mimeDetectionPolicy: (a: number) => [number, number];
     readonly wasmfileextractionconfig_ner: (a: number) => number;
-    readonly wasmfileextractionconfig_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number, k: number, l: number, m: number, n: number, o: number, p: number, q: number, r: number, s: number, t: number, u: number, v: number, w: number, x: number, y: number, z: bigint, a1: number, b1: number, c1: number, d1: number, e1: number, f1: number, g1: number, h1: number, i1: number, j1: number) => number;
+    readonly wasmfileextractionconfig_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number, k: number, l: number, m: number, n: number, o: number, p: number, q: number, r: number, s: number, t: number, u: number, v: number, w: number, x: number, y: number, z: number, a1: bigint, b1: number, c1: number, d1: number, e1: number, f1: number, g1: number, h1: number, i1: number, j1: number, k1: number) => number;
     readonly wasmfileextractionconfig_ocr: (a: number) => number;
     readonly wasmfileextractionconfig_ocrStrategy: (a: number) => any;
     readonly wasmfileextractionconfig_outputFormat: (a: number) => any;
@@ -13450,6 +13500,7 @@ export interface InitOutput {
     readonly wasmfileextractionconfig_set_qrCodes: (a: number, b: number) => void;
     readonly wasmfileextractionconfig_set_redaction: (a: number, b: number) => void;
     readonly wasmfileextractionconfig_set_resultFormat: (a: number, b: number) => void;
+    readonly wasmfileextractionconfig_set_strictParse: (a: number, b: number) => void;
     readonly wasmfileextractionconfig_set_structuredExtraction: (a: number, b: number) => void;
     readonly wasmfileextractionconfig_set_summarization: (a: number, b: number) => void;
     readonly wasmfileextractionconfig_set_timeoutSecs: (a: number, b: number, c: bigint) => void;
@@ -13457,6 +13508,7 @@ export interface InitOutput {
     readonly wasmfileextractionconfig_set_transcription: (a: number, b: number) => void;
     readonly wasmfileextractionconfig_set_translation: (a: number, b: number) => void;
     readonly wasmfileextractionconfig_set_url: (a: number, b: number) => void;
+    readonly wasmfileextractionconfig_strictParse: (a: number) => number;
     readonly wasmfileextractionconfig_structuredExtraction: (a: number) => number;
     readonly wasmfileextractionconfig_summarization: (a: number) => number;
     readonly wasmfileextractionconfig_timeoutSecs: (a: number) => [number, bigint];
@@ -14460,6 +14512,15 @@ export interface InitOutput {
     readonly wasmocrmetadata_tableCols: (a: number) => number;
     readonly wasmocrmetadata_tableCount: (a: number) => number;
     readonly wasmocrmetadata_tableRows: (a: number) => number;
+    readonly wasmocrpagefailure_copyForTransfer: (a: number) => number;
+    readonly wasmocrpagefailure_default: () => number;
+    readonly wasmocrpagefailure_error: (a: number) => [number, number];
+    readonly wasmocrpagefailure_new: (a: number, b: number, c: number, d: number) => number;
+    readonly wasmocrpagefailure_page: (a: number) => number;
+    readonly wasmocrpagefailure_recovered: (a: number) => number;
+    readonly wasmocrpagefailure_set_error: (a: number, b: number, c: number) => void;
+    readonly wasmocrpagefailure_set_page: (a: number, b: number) => void;
+    readonly wasmocrpagefailure_set_recovered: (a: number, b: number) => void;
     readonly wasmocrpipelineconfig_copyForTransfer: (a: number) => number;
     readonly wasmocrpipelineconfig_default: () => number;
     readonly wasmocrpipelineconfig_new: (a: number, b: number, c: number) => number;
@@ -14668,10 +14729,12 @@ export interface InitOutput {
     readonly wasmpageconfig_default: () => number;
     readonly wasmpageconfig_extractPages: (a: number) => number;
     readonly wasmpageconfig_insertPageMarkers: (a: number) => number;
+    readonly wasmpageconfig_keepNativeContent: (a: number) => number;
     readonly wasmpageconfig_markerFormat: (a: number) => [number, number];
-    readonly wasmpageconfig_new: (a: number, b: number, c: number, d: number) => number;
+    readonly wasmpageconfig_new: (a: number, b: number, c: number, d: number, e: number) => number;
     readonly wasmpageconfig_set_extractPages: (a: number, b: number) => void;
     readonly wasmpageconfig_set_insertPageMarkers: (a: number, b: number) => void;
+    readonly wasmpageconfig_set_keepNativeContent: (a: number, b: number) => void;
     readonly wasmpageconfig_set_markerFormat: (a: number, b: number, c: number) => void;
     readonly wasmpagecontent_clearHierarchy: (a: number) => void;
     readonly wasmpagecontent_clearImagePreprocessing: (a: number) => void;
@@ -14684,7 +14747,8 @@ export interface InitOutput {
     readonly wasmpagecontent_imagePreprocessing: (a: number) => number;
     readonly wasmpagecontent_isBlank: (a: number) => number;
     readonly wasmpagecontent_layoutRegions: (a: number) => any;
-    readonly wasmpagecontent_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number, k: number, l: number, m: number, n: number, o: number, p: number, q: number, r: number, s: number) => number;
+    readonly wasmpagecontent_nativeContent: (a: number) => [number, number];
+    readonly wasmpagecontent_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number, k: number, l: number, m: number, n: number, o: number, p: number, q: number, r: number, s: number, t: number, u: number) => number;
     readonly wasmpagecontent_ocrConfidence: (a: number) => number;
     readonly wasmpagecontent_pageNumber: (a: number) => number;
     readonly wasmpagecontent_sectionName: (a: number) => [number, number];
@@ -14694,6 +14758,7 @@ export interface InitOutput {
     readonly wasmpagecontent_set_imagePreprocessing: (a: number, b: number) => void;
     readonly wasmpagecontent_set_isBlank: (a: number, b: number) => void;
     readonly wasmpagecontent_set_layoutRegions: (a: number, b: number, c: number) => void;
+    readonly wasmpagecontent_set_nativeContent: (a: number, b: number, c: number) => void;
     readonly wasmpagecontent_set_ocrConfidence: (a: number, b: number) => void;
     readonly wasmpagecontent_set_pageNumber: (a: number, b: number) => void;
     readonly wasmpagecontent_set_sectionName: (a: number, b: number, c: number) => void;
@@ -15406,10 +15471,6 @@ export interface InitOutput {
     readonly wasmssrfpolicy_allowlist: (a: number) => any;
     readonly wasmssrfpolicy_copyForTransfer: (a: number) => number;
     readonly wasmssrfpolicy_default: () => number;
-    readonly wasmssrfpolicy_denyPrivate: (a: number) => number;
-    readonly wasmssrfpolicy_denylist: (a: number) => any;
-    readonly wasmssrfpolicy_maxRedirects: (a: number) => number;
-    readonly wasmssrfpolicy_new: (a: number, b: number, c: number, d: number, e: number, f: number) => number;
     readonly qcms_profile_precache_output_transform: (a: number) => void;
     readonly lut_inverse_interp16: (a: number, b: number, c: number) => number;
     readonly lut_interp_linear16: (a: number, b: number, c: number) => number;
@@ -15422,9 +15483,13 @@ export interface InitOutput {
     readonly qcms_transform_data_bgra_out_lut_precache: (a: number, b: number, c: number, d: number) => void;
     readonly qcms_profile_is_bogus: (a: number) => number;
     readonly qcms_transform_release: (a: number) => void;
+    readonly wasmssrfpolicy_denyPrivate: (a: number) => number;
+    readonly wasmssrfpolicy_denylist: (a: number) => any;
+    readonly __cxa_atexit: (a: number, b: number, c: number) => number;
+    readonly wasmssrfpolicy_maxRedirects: (a: number) => number;
+    readonly wasmssrfpolicy_new: (a: number, b: number, c: number, d: number, e: number, f: number) => number;
     readonly wasmssrfpolicy_schemeAllowlist: (a: number) => [number, number];
     readonly wasmssrfpolicy_set_allowlist: (a: number, b: any) => void;
-    readonly __cxa_atexit: (a: number, b: number, c: number) => number;
     readonly wasmssrfpolicy_set_denyPrivate: (a: number, b: number) => void;
     readonly wasmssrfpolicy_set_denylist: (a: number, b: any) => void;
     readonly wasmssrfpolicy_set_maxRedirects: (a: number, b: number) => void;
@@ -15801,9 +15866,9 @@ export interface InitOutput {
     readonly xbergengine_new: (a: any, b: any) => [number, number, number];
     readonly qcms_enable_iccv4: () => void;
     readonly xbergengine_ocr: (a: number, b: number, c: number, d: any) => any;
-    readonly wasm_bindgen__convert__closures_____invoke__h1e71e1507830a137: (a: number, b: number, c: any, d: any) => void;
-    readonly wasm_bindgen__convert__closures_____invoke__hb42d2252c7df56c6: (a: number, b: number, c: any) => [number, number];
-    readonly wasm_bindgen__convert__closures_____invoke__he878e50fc99f4ddb: (a: number, b: number) => void;
+    readonly wasm_bindgen__convert__closures_____invoke__h1697dacc3af5e769: (a: number, b: number, c: any, d: any) => void;
+    readonly wasm_bindgen__convert__closures_____invoke__hcbfea0afcfcd2f5c: (a: number, b: number, c: any) => [number, number];
+    readonly wasm_bindgen__convert__closures_____invoke__h697ad9e8f3481d94: (a: number, b: number) => void;
     readonly __wbindgen_malloc_command_export: (a: number, b: number) => number;
     readonly __wbindgen_realloc_command_export: (a: number, b: number, c: number, d: number) => number;
     readonly __wbindgen_exn_store_command_export: (a: number) => void;

package/pkg/web/xberg_wasm.js

@@ -14034,7 +14034,7 @@ export class WasmExtractedDocument {
      * of those handles is dead once this returns. Passing a freshly built value is fine. To
      * keep a handle you already hold, build with `default()` and assign the property instead:
      * the generated setter borrows and leaves your handle alive.
-     * Takes ownership of every element in `tables`, `detectedLanguageConfidences`, `chunks`, `images`, `pages`, `elements`, `ocrElements`, `extractedKeywords`, `processingWarnings`, `annotations`, `children`, `uris`, `revisions`, `llmUsage`, `entities`, `pageClassifications`, `formulas`, `formFields`.
+     * Takes ownership of every element in `tables`, `detectedLanguageConfidences`, `chunks`, `images`, `pages`, `elements`, `ocrElements`, `extractedKeywords`, `processingWarnings`, `annotations`, `children`, `uris`, `revisions`, `llmUsage`, `entities`, `pageClassifications`, `formulas`, `formFields`, `ocrPageFailures`.
      * wasm-bindgen cannot borrow class values nested in an array. To keep those handles,
      * copy each retained element with `copyForTransfer()` before passing the array.
      * @param {string | null} [content]
@@ -14069,8 +14069,9 @@ export class WasmExtractedDocument {
      * @param {WasmTranslation | null} [translation]
      * @param {WasmPageClassification[] | null} [pageClassifications]
      * @param {WasmRedactionReport | null} [redactionReport]
+     * @param {WasmOcrPageFailure[] | null} [ocrPageFailures]
      */
-    constructor(content, mimeType, metadata, tables, counts, processingWarnings, formulas, formFields, extractionMethod, detectedLanguages, detectedLanguageConfidences, chunks, images, pages, elements, djotContent, ocrElements, document, extractedKeywords, qualityScore, annotations, children, uris, revisions, structuredOutput, llmUsage, entities, summary, extractionConfidence, translation, pageClassifications, redactionReport) {
+    constructor(content, mimeType, metadata, tables, counts, processingWarnings, formulas, formFields, extractionMethod, detectedLanguages, detectedLanguageConfidences, chunks, images, pages, elements, djotContent, ocrElements, document, extractedKeywords, qualityScore, annotations, children, uris, revisions, structuredOutput, llmUsage, entities, summary, extractionConfidence, translation, pageClassifications, redactionReport, ocrPageFailures) {
         var ptr0 = isLikeNone(content) ? 0 : passStringToWasm0(content, wasm.__wbindgen_malloc_command_export, wasm.__wbindgen_realloc_command_export);
         var len0 = WASM_VECTOR_LEN;
         var ptr1 = isLikeNone(mimeType) ? 0 : passStringToWasm0(mimeType, wasm.__wbindgen_malloc_command_export, wasm.__wbindgen_realloc_command_export);
@@ -14153,7 +14154,9 @@ export class WasmExtractedDocument {
             _assertClass(redactionReport, WasmRedactionReport);
             ptr28 = redactionReport.__destroy_into_raw();
         }
-        const ret = wasm.wasmextracteddocument_new(ptr0, len0, ptr1, len1, ptr2, ptr3, len3, ptr4, ptr5, len5, ptr6, len6, ptr7, len7, isLikeNone(extractionMethod) ? 3 : extractionMethod, ptr8, len8, ptr9, len9, ptr10, len10, ptr11, len11, ptr12, len12, ptr13, len13, ptr14, ptr15, len15, ptr16, ptr17, len17, !isLikeNone(qualityScore), isLikeNone(qualityScore) ? 0 : qualityScore, ptr18, len18, ptr19, len19, ptr20, len20, ptr21, len21, isLikeNone(structuredOutput) ? 0 : addToExternrefTable0(structuredOutput), ptr22, len22, ptr23, len23, ptr24, ptr25, ptr26, ptr27, len27, ptr28);
+        var ptr29 = isLikeNone(ocrPageFailures) ? 0 : passArrayJsValueToWasm0(ocrPageFailures, wasm.__wbindgen_malloc_command_export);
+        var len29 = WASM_VECTOR_LEN;
+        const ret = wasm.wasmextracteddocument_new(ptr0, len0, ptr1, len1, ptr2, ptr3, len3, ptr4, ptr5, len5, ptr6, len6, ptr7, len7, isLikeNone(extractionMethod) ? 3 : extractionMethod, ptr8, len8, ptr9, len9, ptr10, len10, ptr11, len11, ptr12, len12, ptr13, len13, ptr14, ptr15, len15, ptr16, ptr17, len17, !isLikeNone(qualityScore), isLikeNone(qualityScore) ? 0 : qualityScore, ptr18, len18, ptr19, len19, ptr20, len20, ptr21, len21, isLikeNone(structuredOutput) ? 0 : addToExternrefTable0(structuredOutput), ptr22, len22, ptr23, len23, ptr24, ptr25, ptr26, ptr27, len27, ptr28, ptr29, len29);
         this.__wbg_ptr = ret;
         WasmExtractedDocumentFinalization.register(this, this.__wbg_ptr, this);
         return this;
@@ -14169,6 +14172,17 @@ export class WasmExtractedDocument {
         const ret = wasm.wasmextracteddocument_ocrElements(this.__wbg_ptr);
         return ret;
     }
+    /**
+     * Returns detached copies of the elements in `ocrPageFailures`.
+     *
+     * Changes to the returned values do not update this object. Modify the copies,
+     * then assign the array back to `ocrPageFailures`.
+     * @returns {Array<any> | undefined}
+     */
+    get ocrPageFailures() {
+        const ret = wasm.wasmextracteddocument_ocrPageFailures(this.__wbg_ptr);
+        return ret;
+    }
     /**
      * Returns detached copies of the elements in `pageClassifications`.
      *
@@ -14430,6 +14444,17 @@ export class WasmExtractedDocument {
         var len0 = WASM_VECTOR_LEN;
         wasm.wasmextracteddocument_set_ocrElements(this.__wbg_ptr, ptr0, len0);
     }
+    /**
+     * Takes ownership of every element in `ocrPageFailures`.
+     * wasm-bindgen cannot borrow class values nested in an array; copy each retained
+     * element with `copyForTransfer()` before passing the array.
+     * @param {WasmOcrPageFailure[] | null} [value]
+     */
+    set ocrPageFailures(value) {
+        var ptr0 = isLikeNone(value) ? 0 : passArrayJsValueToWasm0(value, wasm.__wbindgen_malloc_command_export);
+        var len0 = WASM_VECTOR_LEN;
+        wasm.wasmextracteddocument_set_ocrPageFailures(this.__wbg_ptr, ptr0, len0);
+    }
     /**
      * Takes ownership of every element in `pageClassifications`.
      * wasm-bindgen cannot borrow class values nested in an array; copy each retained
@@ -15828,6 +15853,7 @@ export class WasmExtractionConfig {
      * @param {boolean | null} [applyNotebookCellTags]
      * @param {boolean | null} [useLayoutForMarkdown]
      * @param {boolean | null} [includeDocumentStructure]
+     * @param {boolean | null} [strictParse]
      * @param {WasmUrlExtractionConfig | null} [url]
      * @param {number | null} [maxArchiveDepth]
      * @param {WasmOcrConfig | null} [ocr]
@@ -15869,7 +15895,7 @@ export class WasmExtractionConfig {
      * @param {WasmCaptioningConfig | null} [captioning]
      * @param {boolean | null} [qrCodes]
      */
-    constructor(mimeDetectionPolicy, useCache, enableQualityProcessing, forceOcr, ocrStrategy, disableOcr, resultFormat, outputFormat, escapeMarkdown, tableAnchors, jupyterCellRendering, applyNotebookCellTags, useLayoutForMarkdown, includeDocumentStructure, url, maxArchiveDepth, ocr, forceOcrPages, ocrNearEmptyFallback, ocrScannedPageQualityGate, ocrEmbeddedImages, chunking, contentFilter, images, pdfOptions, tokenReduction, languageDetection, pages, keywords, postprocessor, htmlOptions, htmlOutput, extractionTimeoutSecs, maxConcurrentExtractions, securityLimits, maxEmbeddedFileBytes, layout, transcription, acceleration, cacheNamespace, cacheTtlSecs, email, csv, geojson, concurrency, structuredExtraction, ner, redaction, summarization, translation, pageClassification, chunkClassification, captioning, qrCodes) {
+    constructor(mimeDetectionPolicy, useCache, enableQualityProcessing, forceOcr, ocrStrategy, disableOcr, resultFormat, outputFormat, escapeMarkdown, tableAnchors, jupyterCellRendering, applyNotebookCellTags, useLayoutForMarkdown, includeDocumentStructure, strictParse, url, maxArchiveDepth, ocr, forceOcrPages, ocrNearEmptyFallback, ocrScannedPageQualityGate, ocrEmbeddedImages, chunking, contentFilter, images, pdfOptions, tokenReduction, languageDetection, pages, keywords, postprocessor, htmlOptions, htmlOutput, extractionTimeoutSecs, maxConcurrentExtractions, securityLimits, maxEmbeddedFileBytes, layout, transcription, acceleration, cacheNamespace, cacheTtlSecs, email, csv, geojson, concurrency, structuredExtraction, ner, redaction, summarization, translation, pageClassification, chunkClassification, captioning, qrCodes) {
         let ptr0 = 0;
         if (!isLikeNone(url)) {
             _assertClass(url, WasmUrlExtractionConfig);
@@ -16019,7 +16045,7 @@ export class WasmExtractionConfig {
             _assertClass(captioning, WasmCaptioningConfig);
             ptr30 = captioning.__destroy_into_raw();
         }
-        const ret = wasm.wasmextractionconfig_new(isLikeNone(mimeDetectionPolicy) ? 3 : mimeDetectionPolicy, isLikeNone(useCache) ? 0xFFFFFF : useCache ? 1 : 0, isLikeNone(enableQualityProcessing) ? 0xFFFFFF : enableQualityProcessing ? 1 : 0, isLikeNone(forceOcr) ? 0xFFFFFF : forceOcr ? 1 : 0, isLikeNone(ocrStrategy) ? 0 : addToExternrefTable0(ocrStrategy), isLikeNone(disableOcr) ? 0xFFFFFF : disableOcr ? 1 : 0, isLikeNone(resultFormat) ? 2 : resultFormat, isLikeNone(outputFormat) ? 0 : addToExternrefTable0(outputFormat), isLikeNone(escapeMarkdown) ? 0xFFFFFF : escapeMarkdown ? 1 : 0, isLikeNone(tableAnchors) ? 0xFFFFFF : tableAnchors ? 1 : 0, isLikeNone(jupyterCellRendering) ? 3 : jupyterCellRendering, isLikeNone(applyNotebookCellTags) ? 0xFFFFFF : applyNotebookCellTags ? 1 : 0, isLikeNone(useLayoutForMarkdown) ? 0xFFFFFF : useLayoutForMarkdown ? 1 : 0, isLikeNone(includeDocumentStructure) ? 0xFFFFFF : includeDocumentStructure ? 1 : 0, ptr0, isLikeNone(maxArchiveDepth) ? Number.MAX_SAFE_INTEGER : (maxArchiveDepth) >>> 0, ptr1, ptr2, len2, isLikeNone(ocrNearEmptyFallback) ? 0xFFFFFF : ocrNearEmptyFallback ? 1 : 0, isLikeNone(ocrScannedPageQualityGate) ? 0xFFFFFF : ocrScannedPageQualityGate ? 1 : 0, isLikeNone(ocrEmbeddedImages) ? 0xFFFFFF : ocrEmbeddedImages ? 1 : 0, ptr3, ptr4, ptr5, ptr6, ptr7, ptr8, ptr9, ptr10, ptr11, ptr12, ptr13, !isLikeNone(extractionTimeoutSecs), isLikeNone(extractionTimeoutSecs) ? BigInt(0) : extractionTimeoutSecs, isLikeNone(maxConcurrentExtractions) ? Number.MAX_SAFE_INTEGER : (maxConcurrentExtractions) >>> 0, ptr14, !isLikeNone(maxEmbeddedFileBytes), isLikeNone(maxEmbeddedFileBytes) ? BigInt(0) : maxEmbeddedFileBytes, ptr15, ptr16, ptr17, ptr18, len18, !isLikeNone(cacheTtlSecs), isLikeNone(cacheTtlSecs) ? BigInt(0) : cacheTtlSecs, ptr19, ptr20, ptr21, ptr22, ptr23, ptr24, ptr25, ptr26, ptr27, ptr28, ptr29, ptr30, isLikeNone(qrCodes) ? 0xFFFFFF : qrCodes ? 1 : 0);
+        const ret = wasm.wasmextractionconfig_new(isLikeNone(mimeDetectionPolicy) ? 3 : mimeDetectionPolicy, isLikeNone(useCache) ? 0xFFFFFF : useCache ? 1 : 0, isLikeNone(enableQualityProcessing) ? 0xFFFFFF : enableQualityProcessing ? 1 : 0, isLikeNone(forceOcr) ? 0xFFFFFF : forceOcr ? 1 : 0, isLikeNone(ocrStrategy) ? 0 : addToExternrefTable0(ocrStrategy), isLikeNone(disableOcr) ? 0xFFFFFF : disableOcr ? 1 : 0, isLikeNone(resultFormat) ? 2 : resultFormat, isLikeNone(outputFormat) ? 0 : addToExternrefTable0(outputFormat), isLikeNone(escapeMarkdown) ? 0xFFFFFF : escapeMarkdown ? 1 : 0, isLikeNone(tableAnchors) ? 0xFFFFFF : tableAnchors ? 1 : 0, isLikeNone(jupyterCellRendering) ? 3 : jupyterCellRendering, isLikeNone(applyNotebookCellTags) ? 0xFFFFFF : applyNotebookCellTags ? 1 : 0, isLikeNone(useLayoutForMarkdown) ? 0xFFFFFF : useLayoutForMarkdown ? 1 : 0, isLikeNone(includeDocumentStructure) ? 0xFFFFFF : includeDocumentStructure ? 1 : 0, isLikeNone(strictParse) ? 0xFFFFFF : strictParse ? 1 : 0, ptr0, isLikeNone(maxArchiveDepth) ? Number.MAX_SAFE_INTEGER : (maxArchiveDepth) >>> 0, ptr1, ptr2, len2, isLikeNone(ocrNearEmptyFallback) ? 0xFFFFFF : ocrNearEmptyFallback ? 1 : 0, isLikeNone(ocrScannedPageQualityGate) ? 0xFFFFFF : ocrScannedPageQualityGate ? 1 : 0, isLikeNone(ocrEmbeddedImages) ? 0xFFFFFF : ocrEmbeddedImages ? 1 : 0, ptr3, ptr4, ptr5, ptr6, ptr7, ptr8, ptr9, ptr10, ptr11, ptr12, ptr13, !isLikeNone(extractionTimeoutSecs), isLikeNone(extractionTimeoutSecs) ? BigInt(0) : extractionTimeoutSecs, isLikeNone(maxConcurrentExtractions) ? Number.MAX_SAFE_INTEGER : (maxConcurrentExtractions) >>> 0, ptr14, !isLikeNone(maxEmbeddedFileBytes), isLikeNone(maxEmbeddedFileBytes) ? BigInt(0) : maxEmbeddedFileBytes, ptr15, ptr16, ptr17, ptr18, len18, !isLikeNone(cacheTtlSecs), isLikeNone(cacheTtlSecs) ? BigInt(0) : cacheTtlSecs, ptr19, ptr20, ptr21, ptr22, ptr23, ptr24, ptr25, ptr26, ptr27, ptr28, ptr29, ptr30, isLikeNone(qrCodes) ? 0xFFFFFF : qrCodes ? 1 : 0);
         this.__wbg_ptr = ret;
         WasmExtractionConfigFinalization.register(this, this.__wbg_ptr, this);
         return this;
@@ -16482,6 +16508,12 @@ export class WasmExtractionConfig {
         _assertClass(value, WasmSecurityLimits);
         wasm.wasmextractionconfig_set_securityLimits(this.__wbg_ptr, value.__wbg_ptr);
     }
+    /**
+     * @param {boolean} value
+     */
+    set strictParse(value) {
+        wasm.wasmextractionconfig_set_strictParse(this.__wbg_ptr, value);
+    }
     /**
      * @param {WasmStructuredExtractionConfig} value
      */
@@ -16542,6 +16574,13 @@ export class WasmExtractionConfig {
     set useLayoutForMarkdown(value) {
         wasm.wasmextractionconfig_set_useLayoutForMarkdown(this.__wbg_ptr, value);
     }
+    /**
+     * @returns {boolean}
+     */
+    get strictParse() {
+        const ret = wasm.wasmextractionconfig_strictParse(this.__wbg_ptr);
+        return ret !== 0;
+    }
     /**
      * Returns a detached copy of `structuredExtraction`.
      *
@@ -17623,6 +17662,7 @@ export class WasmFileExtractionConfig {
      * @param {WasmResultFormat | null} [resultFormat]
      * @param {any | null} [outputFormat]
      * @param {boolean | null} [includeDocumentStructure]
+     * @param {boolean | null} [strictParse]
      * @param {WasmLayoutDetectionConfig | null} [layout]
      * @param {WasmTranscriptionConfig | null} [transcription]
      * @param {bigint | null} [timeoutSecs]
@@ -17637,7 +17677,7 @@ export class WasmFileExtractionConfig {
      * @param {WasmCaptioningConfig | null} [captioning]
      * @param {boolean | null} [qrCodes]
      */
-    constructor(mimeDetectionPolicy, enableQualityProcessing, ocr, forceOcr, ocrStrategy, forceOcrPages, disableOcr, chunking, contentFilter, images, pdfOptions, tokenReduction, languageDetection, pages, keywords, postprocessor, htmlOptions, htmlOutput, resultFormat, outputFormat, includeDocumentStructure, layout, transcription, timeoutSecs, structuredExtraction, url, ner, redaction, summarization, translation, pageClassification, chunkClassification, captioning, qrCodes) {
+    constructor(mimeDetectionPolicy, enableQualityProcessing, ocr, forceOcr, ocrStrategy, forceOcrPages, disableOcr, chunking, contentFilter, images, pdfOptions, tokenReduction, languageDetection, pages, keywords, postprocessor, htmlOptions, htmlOutput, resultFormat, outputFormat, includeDocumentStructure, strictParse, layout, transcription, timeoutSecs, structuredExtraction, url, ner, redaction, summarization, translation, pageClassification, chunkClassification, captioning, qrCodes) {
         let ptr0 = 0;
         if (!isLikeNone(ocr)) {
             _assertClass(ocr, WasmOcrConfig);
@@ -17755,7 +17795,7 @@ export class WasmFileExtractionConfig {
             _assertClass(captioning, WasmCaptioningConfig);
             ptr23 = captioning.__destroy_into_raw();
         }
-        const ret = wasm.wasmfileextractionconfig_new(isLikeNone(mimeDetectionPolicy) ? 3 : mimeDetectionPolicy, isLikeNone(enableQualityProcessing) ? 0xFFFFFF : enableQualityProcessing ? 1 : 0, ptr0, isLikeNone(forceOcr) ? 0xFFFFFF : forceOcr ? 1 : 0, isLikeNone(ocrStrategy) ? 0 : addToExternrefTable0(ocrStrategy), ptr1, len1, isLikeNone(disableOcr) ? 0xFFFFFF : disableOcr ? 1 : 0, ptr2, ptr3, ptr4, ptr5, ptr6, ptr7, ptr8, ptr9, ptr10, ptr11, ptr12, isLikeNone(resultFormat) ? 2 : resultFormat, isLikeNone(outputFormat) ? 0 : addToExternrefTable0(outputFormat), isLikeNone(includeDocumentStructure) ? 0xFFFFFF : includeDocumentStructure ? 1 : 0, ptr13, ptr14, !isLikeNone(timeoutSecs), isLikeNone(timeoutSecs) ? BigInt(0) : timeoutSecs, ptr15, ptr16, ptr17, ptr18, ptr19, ptr20, ptr21, ptr22, ptr23, isLikeNone(qrCodes) ? 0xFFFFFF : qrCodes ? 1 : 0);
+        const ret = wasm.wasmfileextractionconfig_new(isLikeNone(mimeDetectionPolicy) ? 3 : mimeDetectionPolicy, isLikeNone(enableQualityProcessing) ? 0xFFFFFF : enableQualityProcessing ? 1 : 0, ptr0, isLikeNone(forceOcr) ? 0xFFFFFF : forceOcr ? 1 : 0, isLikeNone(ocrStrategy) ? 0 : addToExternrefTable0(ocrStrategy), ptr1, len1, isLikeNone(disableOcr) ? 0xFFFFFF : disableOcr ? 1 : 0, ptr2, ptr3, ptr4, ptr5, ptr6, ptr7, ptr8, ptr9, ptr10, ptr11, ptr12, isLikeNone(resultFormat) ? 2 : resultFormat, isLikeNone(outputFormat) ? 0 : addToExternrefTable0(outputFormat), isLikeNone(includeDocumentStructure) ? 0xFFFFFF : includeDocumentStructure ? 1 : 0, isLikeNone(strictParse) ? 0xFFFFFF : strictParse ? 1 : 0, ptr13, ptr14, !isLikeNone(timeoutSecs), isLikeNone(timeoutSecs) ? BigInt(0) : timeoutSecs, ptr15, ptr16, ptr17, ptr18, ptr19, ptr20, ptr21, ptr22, ptr23, isLikeNone(qrCodes) ? 0xFFFFFF : qrCodes ? 1 : 0);
         this.__wbg_ptr = ret;
         WasmFileExtractionConfigFinalization.register(this, this.__wbg_ptr, this);
         return this;
@@ -18040,6 +18080,12 @@ export class WasmFileExtractionConfig {
     set resultFormat(value) {
         wasm.wasmfileextractionconfig_set_resultFormat(this.__wbg_ptr, isLikeNone(value) ? 2 : value);
     }
+    /**
+     * @param {boolean | null} [value]
+     */
+    set strictParse(value) {
+        wasm.wasmfileextractionconfig_set_strictParse(this.__wbg_ptr, isLikeNone(value) ? 0xFFFFFF : value ? 1 : 0);
+    }
     /**
      * @param {WasmStructuredExtractionConfig} value
      */
@@ -18088,6 +18134,13 @@ export class WasmFileExtractionConfig {
         _assertClass(value, WasmUrlExtractionConfig);
         wasm.wasmfileextractionconfig_set_url(this.__wbg_ptr, value.__wbg_ptr);
     }
+    /**
+     * @returns {boolean | undefined}
+     */
+    get strictParse() {
+        const ret = wasm.wasmfileextractionconfig_strictParse(this.__wbg_ptr);
+        return ret === 0xFFFFFF ? undefined : ret !== 0;
+    }
     /**
      * Returns a detached copy of `structuredExtraction`.
      *
@@ -29543,6 +29596,121 @@ export class WasmOcrMetadata {
 }
 if (Symbol.dispose) WasmOcrMetadata.prototype[Symbol.dispose] = WasmOcrMetadata.prototype.free;
 
+/**
+ * A page whose OCR failed while extraction of the document continued.
+ *
+ * The warning in `ExtractedDocument.processing_warnings` stays available. One warning
+ * can cover several pages: when an automatic OCR run of selected pages fails as a whole,
+ * it adds one warning and one record for each selected page. When the OCR fallback for the
+ * whole document fails, it adds one warning and one record for each page that has no native
+ * text. Repeated failures on one page make one record.
+ */
+export class WasmOcrPageFailure {
+    static __wrap(ptr) {
+        const obj = Object.create(WasmOcrPageFailure.prototype);
+        obj.__wbg_ptr = ptr;
+        WasmOcrPageFailureFinalization.register(obj, obj.__wbg_ptr, obj);
+        return obj;
+    }
+    static __unwrap(jsValue) {
+        if (!(jsValue instanceof WasmOcrPageFailure)) {
+            return 0;
+        }
+        return jsValue.__destroy_into_raw();
+    }
+    __destroy_into_raw() {
+        const ptr = this.__wbg_ptr;
+        this.__wbg_ptr = 0;
+        WasmOcrPageFailureFinalization.unregister(this);
+        return ptr;
+    }
+    free() {
+        const ptr = this.__destroy_into_raw();
+        wasm.__wbg_wasmocrpagefailure_free(ptr, 0);
+    }
+    /**
+     * Return an independent handle for an ownership-taking API.
+     *
+     * Use `copyForTransfer()` for each class value placed in an array when the original
+     * handle must remain usable.
+     * @returns {WasmOcrPageFailure}
+     */
+    copyForTransfer() {
+        const ret = wasm.wasmocrpagefailure_copyForTransfer(this.__wbg_ptr);
+        return WasmOcrPageFailure.__wrap(ret);
+    }
+    /**
+     * @returns {WasmOcrPageFailure}
+     */
+    static default() {
+        const ret = wasm.wasmocrpagefailure_default();
+        return WasmOcrPageFailure.__wrap(ret);
+    }
+    /**
+     * @returns {string}
+     */
+    get error() {
+        let deferred1_0;
+        let deferred1_1;
+        try {
+            const ret = wasm.wasmocrpagefailure_error(this.__wbg_ptr);
+            deferred1_0 = ret[0];
+            deferred1_1 = ret[1];
+            return getStringFromWasm0(ret[0], ret[1]);
+        } finally {
+            wasm.__wbindgen_free_command_export(deferred1_0, deferred1_1, 1);
+        }
+    }
+    /**
+     * @param {number | null} [page]
+     * @param {string | null} [error]
+     * @param {boolean | null} [recovered]
+     */
+    constructor(page, error, recovered) {
+        var ptr0 = isLikeNone(error) ? 0 : passStringToWasm0(error, wasm.__wbindgen_malloc_command_export, wasm.__wbindgen_realloc_command_export);
+        var len0 = WASM_VECTOR_LEN;
+        const ret = wasm.wasmocrpagefailure_new(isLikeNone(page) ? Number.MAX_SAFE_INTEGER : (page) >>> 0, ptr0, len0, isLikeNone(recovered) ? 0xFFFFFF : recovered ? 1 : 0);
+        this.__wbg_ptr = ret;
+        WasmOcrPageFailureFinalization.register(this, this.__wbg_ptr, this);
+        return this;
+    }
+    /**
+     * @returns {number}
+     */
+    get page() {
+        const ret = wasm.wasmocrpagefailure_page(this.__wbg_ptr);
+        return ret >>> 0;
+    }
+    /**
+     * @returns {boolean}
+     */
+    get recovered() {
+        const ret = wasm.wasmocrpagefailure_recovered(this.__wbg_ptr);
+        return ret !== 0;
+    }
+    /**
+     * @param {string} value
+     */
+    set error(value) {
+        const ptr0 = passStringToWasm0(value, wasm.__wbindgen_malloc_command_export, wasm.__wbindgen_realloc_command_export);
+        const len0 = WASM_VECTOR_LEN;
+        wasm.wasmocrpagefailure_set_error(this.__wbg_ptr, ptr0, len0);
+    }
+    /**
+     * @param {number} value
+     */
+    set page(value) {
+        wasm.wasmocrpagefailure_set_page(this.__wbg_ptr, value);
+    }
+    /**
+     * @param {boolean} value
+     */
+    set recovered(value) {
+        wasm.wasmocrpagefailure_set_recovered(this.__wbg_ptr, value);
+    }
+}
+if (Symbol.dispose) WasmOcrPageFailure.prototype[Symbol.dispose] = WasmOcrPageFailure.prototype.free;
+
 /**
  * Multi-backend OCR pipeline with quality-based fallback.
  *
@@ -31765,6 +31933,13 @@ export class WasmPageConfig {
         const ret = wasm.wasmpageconfig_insertPageMarkers(this.__wbg_ptr);
         return ret !== 0;
     }
+    /**
+     * @returns {boolean}
+     */
+    get keepNativeContent() {
+        const ret = wasm.wasmpageconfig_keepNativeContent(this.__wbg_ptr);
+        return ret !== 0;
+    }
     /**
      * @returns {string}
      */
@@ -31784,11 +31959,12 @@ export class WasmPageConfig {
      * @param {boolean | null} [extractPages]
      * @param {boolean | null} [insertPageMarkers]
      * @param {string | null} [markerFormat]
+     * @param {boolean | null} [keepNativeContent]
      */
-    constructor(extractPages, insertPageMarkers, markerFormat) {
+    constructor(extractPages, insertPageMarkers, markerFormat, keepNativeContent) {
         var ptr0 = isLikeNone(markerFormat) ? 0 : passStringToWasm0(markerFormat, wasm.__wbindgen_malloc_command_export, wasm.__wbindgen_realloc_command_export);
         var len0 = WASM_VECTOR_LEN;
-        const ret = wasm.wasmpageconfig_new(isLikeNone(extractPages) ? 0xFFFFFF : extractPages ? 1 : 0, isLikeNone(insertPageMarkers) ? 0xFFFFFF : insertPageMarkers ? 1 : 0, ptr0, len0);
+        const ret = wasm.wasmpageconfig_new(isLikeNone(extractPages) ? 0xFFFFFF : extractPages ? 1 : 0, isLikeNone(insertPageMarkers) ? 0xFFFFFF : insertPageMarkers ? 1 : 0, ptr0, len0, isLikeNone(keepNativeContent) ? 0xFFFFFF : keepNativeContent ? 1 : 0);
         this.__wbg_ptr = ret;
         WasmPageConfigFinalization.register(this, this.__wbg_ptr, this);
         return this;
@@ -31805,6 +31981,12 @@ export class WasmPageConfig {
     set insertPageMarkers(value) {
         wasm.wasmpageconfig_set_insertPageMarkers(this.__wbg_ptr, value);
     }
+    /**
+     * @param {boolean} value
+     */
+    set keepNativeContent(value) {
+        wasm.wasmpageconfig_set_keepNativeContent(this.__wbg_ptr, value);
+    }
     /**
      * @param {string} value
      */
@@ -31946,6 +32128,18 @@ export class WasmPageContent {
         const ret = wasm.wasmpagecontent_layoutRegions(this.__wbg_ptr);
         return ret;
     }
+    /**
+     * @returns {string | undefined}
+     */
+    get nativeContent() {
+        const ret = wasm.wasmpagecontent_nativeContent(this.__wbg_ptr);
+        let v1;
+        if (ret[0] !== 0) {
+            v1 = getStringFromWasm0(ret[0], ret[1]);
+            wasm.__wbindgen_free_command_export(ret[0], ret[1] * 1, 1);
+        }
+        return v1;
+    }
     /**
      * Takes ownership of `imagePreprocessing`, `hierarchy`, `ocrConfidence`.
      * wasm-bindgen lowers a by-value class argument through `__destroy_into_raw()`, so each
@@ -31967,8 +32161,9 @@ export class WasmPageContent {
      * @param {string | null} [sectionName]
      * @param {string | null} [sheetName]
      * @param {WasmPageOcrConfidence | null} [ocrConfidence]
+     * @param {string | null} [nativeContent]
      */
-    constructor(pageNumber, content, tables, imageIndices, imagePreprocessing, hierarchy, isBlank, layoutRegions, speakerNotes, sectionName, sheetName, ocrConfidence) {
+    constructor(pageNumber, content, tables, imageIndices, imagePreprocessing, hierarchy, isBlank, layoutRegions, speakerNotes, sectionName, sheetName, ocrConfidence, nativeContent) {
         const ptr0 = passStringToWasm0(content, wasm.__wbindgen_malloc_command_export, wasm.__wbindgen_realloc_command_export);
         const len0 = WASM_VECTOR_LEN;
         const ptr1 = passArrayJsValueToWasm0(tables, wasm.__wbindgen_malloc_command_export);
@@ -31998,7 +32193,9 @@ export class WasmPageContent {
             _assertClass(ocrConfidence, WasmPageOcrConfidence);
             ptr9 = ocrConfidence.__destroy_into_raw();
         }
-        const ret = wasm.wasmpagecontent_new(pageNumber, ptr0, len0, ptr1, len1, ptr2, len2, ptr3, ptr4, isLikeNone(isBlank) ? 0xFFFFFF : isBlank ? 1 : 0, ptr5, len5, ptr6, len6, ptr7, len7, ptr8, len8, ptr9);
+        var ptr10 = isLikeNone(nativeContent) ? 0 : passStringToWasm0(nativeContent, wasm.__wbindgen_malloc_command_export, wasm.__wbindgen_realloc_command_export);
+        var len10 = WASM_VECTOR_LEN;
+        const ret = wasm.wasmpagecontent_new(pageNumber, ptr0, len0, ptr1, len1, ptr2, len2, ptr3, ptr4, isLikeNone(isBlank) ? 0xFFFFFF : isBlank ? 1 : 0, ptr5, len5, ptr6, len6, ptr7, len7, ptr8, len8, ptr9, ptr10, len10);
         this.__wbg_ptr = ret;
         WasmPageContentFinalization.register(this, this.__wbg_ptr, this);
         return this;
@@ -32080,6 +32277,14 @@ export class WasmPageContent {
         var len0 = WASM_VECTOR_LEN;
         wasm.wasmpagecontent_set_layoutRegions(this.__wbg_ptr, ptr0, len0);
     }
+    /**
+     * @param {string | null} [value]
+     */
+    set nativeContent(value) {
+        var ptr0 = isLikeNone(value) ? 0 : passStringToWasm0(value, wasm.__wbindgen_malloc_command_export, wasm.__wbindgen_realloc_command_export);
+        var len0 = WASM_VECTOR_LEN;
+        wasm.wasmpagecontent_set_nativeContent(this.__wbg_ptr, ptr0, len0);
+    }
     /**
      * @param {WasmPageOcrConfidence} value
      */
@@ -46151,7 +46356,7 @@ function __wbg_get_imports() {
                     const a = state0.a;
                     state0.a = 0;
                     try {
-                        return wasm_bindgen__convert__closures_____invoke__h1e71e1507830a137(a, state0.b, arg0, arg1);
+                        return wasm_bindgen__convert__closures_____invoke__h1697dacc3af5e769(a, state0.b, arg0, arg1);
                     } finally {
                         state0.a = a;
                     }
@@ -46311,7 +46516,7 @@ function __wbg_get_imports() {
             const ret = arg0.value;
             return ret;
         },
-        __wbg_warn_f4fbcce2635edc12: function(arg0, arg1) {
+        __wbg_warn_b222f740e9ddf70d: function(arg0, arg1) {
             console.warn(getStringFromWasm0(arg0, arg1));
         },
         __wbg_wasmarchiveentry_new: function(arg0) {
@@ -46682,6 +46887,14 @@ function __wbg_get_imports() {
             const ret = WasmOcrElement.__unwrap(arg0);
             return ret;
         },
+        __wbg_wasmocrpagefailure_new: function(arg0) {
+            const ret = WasmOcrPageFailure.__wrap(arg0);
+            return ret;
+        },
+        __wbg_wasmocrpagefailure_unwrap: function(arg0) {
+            const ret = WasmOcrPageFailure.__unwrap(arg0);
+            return ret;
+        },
         __wbg_wasmocrpipelinestage_new: function(arg0) {
             const ret = WasmOcrPipelineStage.__wrap(arg0);
             return ret;
@@ -46883,13 +47096,13 @@ function __wbg_get_imports() {
             return ret;
         },
         __wbindgen_generic_0000000000000001: function(arg0, arg1) {
-            // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [Externref], shim_idx: 691, ret: Result(Unit), inner_ret: Some(Result(Unit)) }, mutable: true }) -> Externref`.
-            const ret = makeMutClosure(arg0, arg1, wasm_bindgen__convert__closures_____invoke__hb42d2252c7df56c6);
+            // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [Externref], shim_idx: 684, ret: Result(Unit), inner_ret: Some(Result(Unit)) }, mutable: true }) -> Externref`.
+            const ret = makeMutClosure(arg0, arg1, wasm_bindgen__convert__closures_____invoke__hcbfea0afcfcd2f5c);
             return ret;
         },
         __wbindgen_generic_0000000000000002: function(arg0, arg1) {
-            // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [], shim_idx: 1134, ret: Unit, inner_ret: Some(Unit) }, mutable: true }) -> Externref`.
-            const ret = makeMutClosure(arg0, arg1, wasm_bindgen__convert__closures_____invoke__he878e50fc99f4ddb);
+            // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [], shim_idx: 1131, ret: Unit, inner_ret: Some(Unit) }, mutable: true }) -> Externref`.
+            const ret = makeMutClosure(arg0, arg1, wasm_bindgen__convert__closures_____invoke__h697ad9e8f3481d94);
             return ret;
         },
         __wbindgen_generic_0000000000000003: function(arg0) {
@@ -46957,19 +47170,19 @@ function __wbg_get_imports() {
     };
 }
 
-function wasm_bindgen__convert__closures_____invoke__he878e50fc99f4ddb(arg0, arg1) {
-    wasm.wasm_bindgen__convert__closures_____invoke__he878e50fc99f4ddb(arg0, arg1);
+function wasm_bindgen__convert__closures_____invoke__h697ad9e8f3481d94(arg0, arg1) {
+    wasm.wasm_bindgen__convert__closures_____invoke__h697ad9e8f3481d94(arg0, arg1);
 }
 
-function wasm_bindgen__convert__closures_____invoke__hb42d2252c7df56c6(arg0, arg1, arg2) {
-    const ret = wasm.wasm_bindgen__convert__closures_____invoke__hb42d2252c7df56c6(arg0, arg1, arg2);
+function wasm_bindgen__convert__closures_____invoke__hcbfea0afcfcd2f5c(arg0, arg1, arg2) {
+    const ret = wasm.wasm_bindgen__convert__closures_____invoke__hcbfea0afcfcd2f5c(arg0, arg1, arg2);
     if (ret[1]) {
         throw takeFromExternrefTable0(ret[0]);
     }
 }
 
-function wasm_bindgen__convert__closures_____invoke__h1e71e1507830a137(arg0, arg1, arg2, arg3) {
-    wasm.wasm_bindgen__convert__closures_____invoke__h1e71e1507830a137(arg0, arg1, arg2, arg3);
+function wasm_bindgen__convert__closures_____invoke__h1697dacc3af5e769(arg0, arg1, arg2, arg3) {
+    wasm.wasm_bindgen__convert__closures_____invoke__h1697dacc3af5e769(arg0, arg1, arg2, arg3);
 }
 
 
@@ -47445,6 +47658,9 @@ const WasmOcrExtractionResultFinalization = (typeof FinalizationRegistry === 'un
 const WasmOcrMetadataFinalization = (typeof FinalizationRegistry === 'undefined')
     ? { register: () => {}, unregister: () => {} }
     : new FinalizationRegistry(ptr => wasm.__wbg_wasmocrmetadata_free(ptr, 1));
+const WasmOcrPageFailureFinalization = (typeof FinalizationRegistry === 'undefined')
+    ? { register: () => {}, unregister: () => {} }
+    : new FinalizationRegistry(ptr => wasm.__wbg_wasmocrpagefailure_free(ptr, 1));
 const WasmOcrPipelineConfigFinalization = (typeof FinalizationRegistry === 'undefined')
     ? { register: () => {}, unregister: () => {} }
     : new FinalizationRegistry(ptr => wasm.__wbg_wasmocrpipelineconfig_free(ptr, 1));

package/pkg/web/xberg_wasm_bg.wasm

@@ -1,4 +1,4 @@
 [diffend] Oversized file quarantined before diffing.
 name: package/pkg/web/xberg_wasm_bg.wasm
-size: 37951907 bytes
-sha256: cd631841405fe389502e76d02c3c634e10375ac834fc228098d5e48c3c5d0579
+size: 38155686 bytes
+sha256: 36d9eac1ba5c078a3fe66b091f9385701ed3b0608f1d2db549c990a3d9886001

package/pkg/web/xberg_wasm_bg.wasm.d.ts

@@ -155,6 +155,7 @@ export const __wbg_wasmocrelement_free: (a: number, b: number) => void;
 export const __wbg_wasmocrelementconfig_free: (a: number, b: number) => void;
 export const __wbg_wasmocrextractionresult_free: (a: number, b: number) => void;
 export const __wbg_wasmocrmetadata_free: (a: number, b: number) => void;
+export const __wbg_wasmocrpagefailure_free: (a: number, b: number) => void;
 export const __wbg_wasmocrpipelineconfig_free: (a: number, b: number) => void;
 export const __wbg_wasmocrpipelinestage_free: (a: number, b: number) => void;
 export const __wbg_wasmocrpoint_free: (a: number, b: number) => void;
@@ -1508,8 +1509,9 @@ export const wasmextracteddocument_images: (a: number) => any;
 export const wasmextracteddocument_llmUsage: (a: number) => any;
 export const wasmextracteddocument_metadata: (a: number) => number;
 export const wasmextracteddocument_mimeType: (a: number) => [number, number];
-export const wasmextracteddocument_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number, k: number, l: number, m: number, n: number, o: number, p: number, q: number, r: number, s: number, t: number, u: number, v: number, w: number, x: number, y: number, z: number, a1: number, b1: number, c1: number, d1: number, e1: number, f1: number, g1: number, h1: number, i1: number, j1: number, k1: number, l1: number, m1: number, n1: number, o1: number, p1: number, q1: number, r1: number, s1: number, t1: number, u1: number, v1: number, w1: number, x1: number, y1: number, z1: number, a2: number, b2: number) => number;
+export const wasmextracteddocument_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number, k: number, l: number, m: number, n: number, o: number, p: number, q: number, r: number, s: number, t: number, u: number, v: number, w: number, x: number, y: number, z: number, a1: number, b1: number, c1: number, d1: number, e1: number, f1: number, g1: number, h1: number, i1: number, j1: number, k1: number, l1: number, m1: number, n1: number, o1: number, p1: number, q1: number, r1: number, s1: number, t1: number, u1: number, v1: number, w1: number, x1: number, y1: number, z1: number, a2: number, b2: number, c2: number, d2: number) => number;
 export const wasmextracteddocument_ocrElements: (a: number) => any;
+export const wasmextracteddocument_ocrPageFailures: (a: number) => any;
 export const wasmextracteddocument_pageClassifications: (a: number) => any;
 export const wasmextracteddocument_pages: (a: number) => any;
 export const wasmextracteddocument_processingWarnings: (a: number) => [number, number];
@@ -1537,6 +1539,7 @@ export const wasmextracteddocument_set_llmUsage: (a: number, b: number, c: numbe
 export const wasmextracteddocument_set_metadata: (a: number, b: number) => void;
 export const wasmextracteddocument_set_mimeType: (a: number, b: number, c: number) => void;
 export const wasmextracteddocument_set_ocrElements: (a: number, b: number, c: number) => void;
+export const wasmextracteddocument_set_ocrPageFailures: (a: number, b: number, c: number) => void;
 export const wasmextracteddocument_set_pageClassifications: (a: number, b: number, c: number) => void;
 export const wasmextracteddocument_set_pages: (a: number, b: number, c: number) => void;
 export const wasmextracteddocument_set_processingWarnings: (a: number, b: number, c: number) => void;
@@ -1702,7 +1705,7 @@ export const wasmextractionconfig_mimeDetectionPolicy: (a: number) => [number, n
 export const wasmextractionconfig_needsImageData: (a: number) => number;
 export const wasmextractionconfig_needsImageProcessing: (a: number) => number;
 export const wasmextractionconfig_ner: (a: number) => number;
-export const wasmextractionconfig_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number, k: number, l: number, m: number, n: number, o: number, p: number, q: number, r: number, s: number, t: number, u: number, v: number, w: number, x: number, y: number, z: number, a1: number, b1: number, c1: number, d1: number, e1: number, f1: number, g1: number, h1: number, i1: bigint, j1: number, k1: number, l1: number, m1: bigint, n1: number, o1: number, p1: number, q1: number, r1: number, s1: number, t1: bigint, u1: number, v1: number, w1: number, x1: number, y1: number, z1: number, a2: number, b2: number, c2: number, d2: number, e2: number, f2: number, g2: number) => number;
+export const wasmextractionconfig_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number, k: number, l: number, m: number, n: number, o: number, p: number, q: number, r: number, s: number, t: number, u: number, v: number, w: number, x: number, y: number, z: number, a1: number, b1: number, c1: number, d1: number, e1: number, f1: number, g1: number, h1: number, i1: number, j1: bigint, k1: number, l1: number, m1: number, n1: bigint, o1: number, p1: number, q1: number, r1: number, s1: number, t1: number, u1: bigint, v1: number, w1: number, x1: number, y1: number, z1: number, a2: number, b2: number, c2: number, d2: number, e2: number, f2: number, g2: number, h2: number) => number;
 export const wasmextractionconfig_ocr: (a: number) => number;
 export const wasmextractionconfig_ocrEmbeddedImages: (a: number) => number;
 export const wasmextractionconfig_ocrNearEmptyFallback: (a: number) => number;
@@ -1763,6 +1766,7 @@ export const wasmextractionconfig_set_qrCodes: (a: number, b: number) => void;
 export const wasmextractionconfig_set_redaction: (a: number, b: number) => void;
 export const wasmextractionconfig_set_resultFormat: (a: number, b: number) => void;
 export const wasmextractionconfig_set_securityLimits: (a: number, b: number) => void;
+export const wasmextractionconfig_set_strictParse: (a: number, b: number) => void;
 export const wasmextractionconfig_set_structuredExtraction: (a: number, b: number) => void;
 export const wasmextractionconfig_set_summarization: (a: number, b: number) => void;
 export const wasmextractionconfig_set_tableAnchors: (a: number, b: number) => void;
@@ -1772,6 +1776,7 @@ export const wasmextractionconfig_set_translation: (a: number, b: number) => voi
 export const wasmextractionconfig_set_url: (a: number, b: number) => void;
 export const wasmextractionconfig_set_useCache: (a: number, b: number) => void;
 export const wasmextractionconfig_set_useLayoutForMarkdown: (a: number, b: number) => void;
+export const wasmextractionconfig_strictParse: (a: number) => number;
 export const wasmextractionconfig_structuredExtraction: (a: number) => number;
 export const wasmextractionconfig_summarization: (a: number) => number;
 export const wasmextractionconfig_tableAnchors: (a: number) => number;
@@ -1878,7 +1883,7 @@ export const wasmfileextractionconfig_languageDetection: (a: number) => number;
 export const wasmfileextractionconfig_layout: (a: number) => number;
 export const wasmfileextractionconfig_mimeDetectionPolicy: (a: number) => [number, number];
 export const wasmfileextractionconfig_ner: (a: number) => number;
-export const wasmfileextractionconfig_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number, k: number, l: number, m: number, n: number, o: number, p: number, q: number, r: number, s: number, t: number, u: number, v: number, w: number, x: number, y: number, z: bigint, a1: number, b1: number, c1: number, d1: number, e1: number, f1: number, g1: number, h1: number, i1: number, j1: number) => number;
+export const wasmfileextractionconfig_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number, k: number, l: number, m: number, n: number, o: number, p: number, q: number, r: number, s: number, t: number, u: number, v: number, w: number, x: number, y: number, z: number, a1: bigint, b1: number, c1: number, d1: number, e1: number, f1: number, g1: number, h1: number, i1: number, j1: number, k1: number) => number;
 export const wasmfileextractionconfig_ocr: (a: number) => number;
 export const wasmfileextractionconfig_ocrStrategy: (a: number) => any;
 export const wasmfileextractionconfig_outputFormat: (a: number) => any;
@@ -1916,6 +1921,7 @@ export const wasmfileextractionconfig_set_postprocessor: (a: number, b: number)
 export const wasmfileextractionconfig_set_qrCodes: (a: number, b: number) => void;
 export const wasmfileextractionconfig_set_redaction: (a: number, b: number) => void;
 export const wasmfileextractionconfig_set_resultFormat: (a: number, b: number) => void;
+export const wasmfileextractionconfig_set_strictParse: (a: number, b: number) => void;
 export const wasmfileextractionconfig_set_structuredExtraction: (a: number, b: number) => void;
 export const wasmfileextractionconfig_set_summarization: (a: number, b: number) => void;
 export const wasmfileextractionconfig_set_timeoutSecs: (a: number, b: number, c: bigint) => void;
@@ -1923,6 +1929,7 @@ export const wasmfileextractionconfig_set_tokenReduction: (a: number, b: number)
 export const wasmfileextractionconfig_set_transcription: (a: number, b: number) => void;
 export const wasmfileextractionconfig_set_translation: (a: number, b: number) => void;
 export const wasmfileextractionconfig_set_url: (a: number, b: number) => void;
+export const wasmfileextractionconfig_strictParse: (a: number) => number;
 export const wasmfileextractionconfig_structuredExtraction: (a: number) => number;
 export const wasmfileextractionconfig_summarization: (a: number) => number;
 export const wasmfileextractionconfig_timeoutSecs: (a: number) => [number, bigint];
@@ -2926,6 +2933,15 @@ export const wasmocrmetadata_set_tableRows: (a: number, b: number) => void;
 export const wasmocrmetadata_tableCols: (a: number) => number;
 export const wasmocrmetadata_tableCount: (a: number) => number;
 export const wasmocrmetadata_tableRows: (a: number) => number;
+export const wasmocrpagefailure_copyForTransfer: (a: number) => number;
+export const wasmocrpagefailure_default: () => number;
+export const wasmocrpagefailure_error: (a: number) => [number, number];
+export const wasmocrpagefailure_new: (a: number, b: number, c: number, d: number) => number;
+export const wasmocrpagefailure_page: (a: number) => number;
+export const wasmocrpagefailure_recovered: (a: number) => number;
+export const wasmocrpagefailure_set_error: (a: number, b: number, c: number) => void;
+export const wasmocrpagefailure_set_page: (a: number, b: number) => void;
+export const wasmocrpagefailure_set_recovered: (a: number, b: number) => void;
 export const wasmocrpipelineconfig_copyForTransfer: (a: number) => number;
 export const wasmocrpipelineconfig_default: () => number;
 export const wasmocrpipelineconfig_new: (a: number, b: number, c: number) => number;
@@ -3134,10 +3150,12 @@ export const wasmpageconfig_copyForTransfer: (a: number) => number;
 export const wasmpageconfig_default: () => number;
 export const wasmpageconfig_extractPages: (a: number) => number;
 export const wasmpageconfig_insertPageMarkers: (a: number) => number;
+export const wasmpageconfig_keepNativeContent: (a: number) => number;
 export const wasmpageconfig_markerFormat: (a: number) => [number, number];
-export const wasmpageconfig_new: (a: number, b: number, c: number, d: number) => number;
+export const wasmpageconfig_new: (a: number, b: number, c: number, d: number, e: number) => number;
 export const wasmpageconfig_set_extractPages: (a: number, b: number) => void;
 export const wasmpageconfig_set_insertPageMarkers: (a: number, b: number) => void;
+export const wasmpageconfig_set_keepNativeContent: (a: number, b: number) => void;
 export const wasmpageconfig_set_markerFormat: (a: number, b: number, c: number) => void;
 export const wasmpagecontent_clearHierarchy: (a: number) => void;
 export const wasmpagecontent_clearImagePreprocessing: (a: number) => void;
@@ -3150,7 +3168,8 @@ export const wasmpagecontent_imageIndices: (a: number) => [number, number];
 export const wasmpagecontent_imagePreprocessing: (a: number) => number;
 export const wasmpagecontent_isBlank: (a: number) => number;
 export const wasmpagecontent_layoutRegions: (a: number) => any;
-export const wasmpagecontent_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number, k: number, l: number, m: number, n: number, o: number, p: number, q: number, r: number, s: number) => number;
+export const wasmpagecontent_nativeContent: (a: number) => [number, number];
+export const wasmpagecontent_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number, k: number, l: number, m: number, n: number, o: number, p: number, q: number, r: number, s: number, t: number, u: number) => number;
 export const wasmpagecontent_ocrConfidence: (a: number) => number;
 export const wasmpagecontent_pageNumber: (a: number) => number;
 export const wasmpagecontent_sectionName: (a: number) => [number, number];
@@ -3160,6 +3179,7 @@ export const wasmpagecontent_set_imageIndices: (a: number, b: number, c: number)
 export const wasmpagecontent_set_imagePreprocessing: (a: number, b: number) => void;
 export const wasmpagecontent_set_isBlank: (a: number, b: number) => void;
 export const wasmpagecontent_set_layoutRegions: (a: number, b: number, c: number) => void;
+export const wasmpagecontent_set_nativeContent: (a: number, b: number, c: number) => void;
 export const wasmpagecontent_set_ocrConfidence: (a: number, b: number) => void;
 export const wasmpagecontent_set_pageNumber: (a: number, b: number) => void;
 export const wasmpagecontent_set_sectionName: (a: number, b: number, c: number) => void;
@@ -3872,10 +3892,6 @@ export const wasmsparseembeddingpreset_set_name: (a: number, b: number, c: numbe
 export const wasmssrfpolicy_allowlist: (a: number) => any;
 export const wasmssrfpolicy_copyForTransfer: (a: number) => number;
 export const wasmssrfpolicy_default: () => number;
-export const wasmssrfpolicy_denyPrivate: (a: number) => number;
-export const wasmssrfpolicy_denylist: (a: number) => any;
-export const wasmssrfpolicy_maxRedirects: (a: number) => number;
-export const wasmssrfpolicy_new: (a: number, b: number, c: number, d: number, e: number, f: number) => number;
 export const qcms_profile_precache_output_transform: (a: number) => void;
 export const lut_inverse_interp16: (a: number, b: number, c: number) => number;
 export const lut_interp_linear16: (a: number, b: number, c: number) => number;
@@ -3888,9 +3904,13 @@ export const qcms_transform_data_rgba_out_lut_precache: (a: number, b: number, c
 export const qcms_transform_data_bgra_out_lut_precache: (a: number, b: number, c: number, d: number) => void;
 export const qcms_profile_is_bogus: (a: number) => number;
 export const qcms_transform_release: (a: number) => void;
+export const wasmssrfpolicy_denyPrivate: (a: number) => number;
+export const wasmssrfpolicy_denylist: (a: number) => any;
+export const __cxa_atexit: (a: number, b: number, c: number) => number;
+export const wasmssrfpolicy_maxRedirects: (a: number) => number;
+export const wasmssrfpolicy_new: (a: number, b: number, c: number, d: number, e: number, f: number) => number;
 export const wasmssrfpolicy_schemeAllowlist: (a: number) => [number, number];
 export const wasmssrfpolicy_set_allowlist: (a: number, b: any) => void;
-export const __cxa_atexit: (a: number, b: number, c: number) => number;
 export const wasmssrfpolicy_set_denyPrivate: (a: number, b: number) => void;
 export const wasmssrfpolicy_set_denylist: (a: number, b: any) => void;
 export const wasmssrfpolicy_set_maxRedirects: (a: number, b: number) => void;
@@ -4267,9 +4287,9 @@ export const xbergengine_ner: (a: number, b: number, c: number, d: any) => any;
 export const xbergengine_new: (a: any, b: any) => [number, number, number];
 export const qcms_enable_iccv4: () => void;
 export const xbergengine_ocr: (a: number, b: number, c: number, d: any) => any;
-export const wasm_bindgen__convert__closures_____invoke__h1e71e1507830a137: (a: number, b: number, c: any, d: any) => void;
-export const wasm_bindgen__convert__closures_____invoke__hb42d2252c7df56c6: (a: number, b: number, c: any) => [number, number];
-export const wasm_bindgen__convert__closures_____invoke__he878e50fc99f4ddb: (a: number, b: number) => void;
+export const wasm_bindgen__convert__closures_____invoke__h1697dacc3af5e769: (a: number, b: number, c: any, d: any) => void;
+export const wasm_bindgen__convert__closures_____invoke__hcbfea0afcfcd2f5c: (a: number, b: number, c: any) => [number, number];
+export const wasm_bindgen__convert__closures_____invoke__h697ad9e8f3481d94: (a: number, b: number) => void;
 export const __wbindgen_malloc_command_export: (a: number, b: number) => number;
 export const __wbindgen_realloc_command_export: (a: number, b: number, c: number, d: number) => number;
 export const __wbindgen_exn_store_command_export: (a: number) => void;