@heripo/pdf-parser 0.1.26 → 0.1.28

package/dist/index.d.cts

@@ -1,9 +1,38 @@
 import { LoggerMethods } from '@heripo/logger';
-import { OcrStrategy, TokenUsageReport } from '@heripo/model';
+import { OcrStrategy, TokenUsageReport, ReviewAssistanceProgressEvent } from '@heripo/model';
+export { ReviewAssistanceProgressEvent, ReviewAssistanceProgressStatus, ReviewAssistanceProgressSubstage } from '@heripo/model';
 import { LanguageModel } from 'ai';
 import { ConversionOptions } from 'docling-sdk';
 import { LLMTokenUsageAggregator } from '@heripo/shared';
 
+interface ReviewAssistanceOptions {
+    /** Enable page-level review assistance. Defaults to false. */
+    enabled?: boolean;
+    /** Minimum final confidence for direct mutation. Defaults to 0.85. */
+    autoApplyThreshold?: number;
+    /** Minimum final confidence for sidecar proposals. Defaults to 0.5. */
+    proposalThreshold?: number;
+    /** Maximum retries per page-level VLM call. Defaults to 3. */
+    maxRetries?: number;
+    /** VLM generation temperature. Defaults to 0. */
+    temperature?: number;
+    /** Language for human-readable review reasons. Defaults to English. */
+    outputLanguage?: string;
+}
+interface NormalizedReviewAssistanceOptions {
+    enabled: boolean;
+    concurrency: number;
+    autoApplyThreshold: number;
+    proposalThreshold: number;
+    maxRetries: number;
+    temperature: number;
+    outputLanguage: string;
+}
+type ReviewAssistanceOptionInput = boolean | ReviewAssistanceOptions | undefined;
+declare const REVIEW_ASSISTANCE_DEFAULTS: NormalizedReviewAssistanceOptions;
+declare function normalizeReviewAssistanceOptions(value: ReviewAssistanceOptionInput): NormalizedReviewAssistanceOptions;
+declare function isReviewAssistanceEnabled(value: ReviewAssistanceOptionInput): boolean;
+
 /**
  * Callback function invoked after PDF conversion completes
  * @param outputPath Absolute path to the output directory containing result files
@@ -25,6 +54,11 @@ type PDFConvertOptions = Omit<ConversionOptions, 'to_formats' | 'image_export_mo
     vlmProcessorModel?: LanguageModel;
     /** Concurrency for VLM page processing (default: 1) */
     vlmConcurrency?: number;
+    /**
+     * Optional page-level review assistance.
+     * Boolean true enables defaults; object form still requires enabled: true.
+     */
+    reviewAssistance?: boolean | ReviewAssistanceOptions;
     /** Skip sampling and default to ocrmac */
     skipSampling?: boolean;
     /** Force a specific OCR method, bypassing sampling */
@@ -33,6 +67,8 @@ type PDFConvertOptions = Omit<ConversionOptions, 'to_formats' | 'image_export_mo
     aggregator?: LLMTokenUsageAggregator;
     /** Callback fired after each batch of VLM pages completes, with cumulative token usage */
     onTokenUsage?: (report: TokenUsageReport) => void;
+    /** Callback fired when Review Assistance advances through parser substages */
+    onReviewAssistanceProgress?: (event: ReviewAssistanceProgressEvent) => void;
     /** Document processing timeout in seconds for the Docling server (default: server default) */
     document_timeout?: number;
     /** Enable chunked conversion for large PDFs (local files only) */
@@ -194,7 +230,8 @@ declare class InvalidDocumentTypeError extends Error {
  * The DoclingDocumentAssembler converts these into a full DoclingDocument.
  */
 /** Allowed element types matching DoclingDocument text labels */
-type VlmElementType = 'text' | 'section_header' | 'caption' | 'footnote' | 'page_header' | 'page_footer' | 'list_item' | 'picture' | 'table';
+declare const VLM_ELEMENT_TYPES: readonly ["text", "section_header", "caption", "footnote", "page_header", "page_footer", "list_item", "picture", "table"];
+type VlmElementType = (typeof VLM_ELEMENT_TYPES)[number];
 /**
  * Normalized bounding box with coordinates in the range 0.0 to 1.0,
  * using top-left origin (standard image coordinates).
@@ -229,7 +266,8 @@ interface VlmPageElement {
     bbox?: VlmBBox;
 }
 /** Types of quality issues detected in VLM responses */
-type VlmQualityIssueType = 'placeholder_text' | 'script_anomaly' | 'meta_description' | 'repetitive_pattern';
+declare const VLM_QUALITY_ISSUE_TYPES: readonly ["placeholder_text", "script_anomaly", "meta_description", "repetitive_pattern"];
+type VlmQualityIssueType = (typeof VLM_QUALITY_ISSUE_TYPES)[number];
 /** Quality metadata for a processed page */
 interface VlmPageQuality {
     /** Whether the page passed quality validation (after possible retry) */
@@ -296,4 +334,4 @@ declare class VlmResponseValidator {
     private static detectRepetitivePattern;
 }
 
-export { type ConversionCompleteCallback, type ConvertWithStrategyResult, ImagePdfFallbackError, InvalidDocumentTypeError, type PDFConvertOptions, PDFParser, type VlmPageQuality, type VlmQualityIssue, type VlmQualityIssueType, VlmResponseValidator, type VlmValidationResult };
+export { type ConversionCompleteCallback, type ConvertWithStrategyResult, ImagePdfFallbackError, InvalidDocumentTypeError, type NormalizedReviewAssistanceOptions, type PDFConvertOptions, PDFParser, REVIEW_ASSISTANCE_DEFAULTS, type ReviewAssistanceOptions, type VlmPageQuality, type VlmQualityIssue, type VlmQualityIssueType, VlmResponseValidator, type VlmValidationResult, isReviewAssistanceEnabled, normalizeReviewAssistanceOptions };

package/dist/index.d.ts

@@ -1,9 +1,38 @@
 import { LoggerMethods } from '@heripo/logger';
-import { OcrStrategy, TokenUsageReport } from '@heripo/model';
+import { OcrStrategy, TokenUsageReport, ReviewAssistanceProgressEvent } from '@heripo/model';
+export { ReviewAssistanceProgressEvent, ReviewAssistanceProgressStatus, ReviewAssistanceProgressSubstage } from '@heripo/model';
 import { LanguageModel } from 'ai';
 import { ConversionOptions } from 'docling-sdk';
 import { LLMTokenUsageAggregator } from '@heripo/shared';
 
+interface ReviewAssistanceOptions {
+    /** Enable page-level review assistance. Defaults to false. */
+    enabled?: boolean;
+    /** Minimum final confidence for direct mutation. Defaults to 0.85. */
+    autoApplyThreshold?: number;
+    /** Minimum final confidence for sidecar proposals. Defaults to 0.5. */
+    proposalThreshold?: number;
+    /** Maximum retries per page-level VLM call. Defaults to 3. */
+    maxRetries?: number;
+    /** VLM generation temperature. Defaults to 0. */
+    temperature?: number;
+    /** Language for human-readable review reasons. Defaults to English. */
+    outputLanguage?: string;
+}
+interface NormalizedReviewAssistanceOptions {
+    enabled: boolean;
+    concurrency: number;
+    autoApplyThreshold: number;
+    proposalThreshold: number;
+    maxRetries: number;
+    temperature: number;
+    outputLanguage: string;
+}
+type ReviewAssistanceOptionInput = boolean | ReviewAssistanceOptions | undefined;
+declare const REVIEW_ASSISTANCE_DEFAULTS: NormalizedReviewAssistanceOptions;
+declare function normalizeReviewAssistanceOptions(value: ReviewAssistanceOptionInput): NormalizedReviewAssistanceOptions;
+declare function isReviewAssistanceEnabled(value: ReviewAssistanceOptionInput): boolean;
+
 /**
  * Callback function invoked after PDF conversion completes
  * @param outputPath Absolute path to the output directory containing result files
@@ -25,6 +54,11 @@ type PDFConvertOptions = Omit<ConversionOptions, 'to_formats' | 'image_export_mo
     vlmProcessorModel?: LanguageModel;
     /** Concurrency for VLM page processing (default: 1) */
     vlmConcurrency?: number;
+    /**
+     * Optional page-level review assistance.
+     * Boolean true enables defaults; object form still requires enabled: true.
+     */
+    reviewAssistance?: boolean | ReviewAssistanceOptions;
     /** Skip sampling and default to ocrmac */
     skipSampling?: boolean;
     /** Force a specific OCR method, bypassing sampling */
@@ -33,6 +67,8 @@ type PDFConvertOptions = Omit<ConversionOptions, 'to_formats' | 'image_export_mo
     aggregator?: LLMTokenUsageAggregator;
     /** Callback fired after each batch of VLM pages completes, with cumulative token usage */
     onTokenUsage?: (report: TokenUsageReport) => void;
+    /** Callback fired when Review Assistance advances through parser substages */
+    onReviewAssistanceProgress?: (event: ReviewAssistanceProgressEvent) => void;
     /** Document processing timeout in seconds for the Docling server (default: server default) */
     document_timeout?: number;
     /** Enable chunked conversion for large PDFs (local files only) */
@@ -194,7 +230,8 @@ declare class InvalidDocumentTypeError extends Error {
  * The DoclingDocumentAssembler converts these into a full DoclingDocument.
  */
 /** Allowed element types matching DoclingDocument text labels */
-type VlmElementType = 'text' | 'section_header' | 'caption' | 'footnote' | 'page_header' | 'page_footer' | 'list_item' | 'picture' | 'table';
+declare const VLM_ELEMENT_TYPES: readonly ["text", "section_header", "caption", "footnote", "page_header", "page_footer", "list_item", "picture", "table"];
+type VlmElementType = (typeof VLM_ELEMENT_TYPES)[number];
 /**
  * Normalized bounding box with coordinates in the range 0.0 to 1.0,
  * using top-left origin (standard image coordinates).
@@ -229,7 +266,8 @@ interface VlmPageElement {
     bbox?: VlmBBox;
 }
 /** Types of quality issues detected in VLM responses */
-type VlmQualityIssueType = 'placeholder_text' | 'script_anomaly' | 'meta_description' | 'repetitive_pattern';
+declare const VLM_QUALITY_ISSUE_TYPES: readonly ["placeholder_text", "script_anomaly", "meta_description", "repetitive_pattern"];
+type VlmQualityIssueType = (typeof VLM_QUALITY_ISSUE_TYPES)[number];
 /** Quality metadata for a processed page */
 interface VlmPageQuality {
     /** Whether the page passed quality validation (after possible retry) */
@@ -296,4 +334,4 @@ declare class VlmResponseValidator {
     private static detectRepetitivePattern;
 }
 
-export { type ConversionCompleteCallback, type ConvertWithStrategyResult, ImagePdfFallbackError, InvalidDocumentTypeError, type PDFConvertOptions, PDFParser, type VlmPageQuality, type VlmQualityIssue, type VlmQualityIssueType, VlmResponseValidator, type VlmValidationResult };
+export { type ConversionCompleteCallback, type ConvertWithStrategyResult, ImagePdfFallbackError, InvalidDocumentTypeError, type NormalizedReviewAssistanceOptions, type PDFConvertOptions, PDFParser, REVIEW_ASSISTANCE_DEFAULTS, type ReviewAssistanceOptions, type VlmPageQuality, type VlmQualityIssue, type VlmQualityIssueType, VlmResponseValidator, type VlmValidationResult, isReviewAssistanceEnabled, normalizeReviewAssistanceOptions };