@heripo/pdf-parser 0.1.34 → 0.1.36

package/dist/index.d.cts

@@ -54,6 +54,12 @@ interface PDFCorrectionOptions {
      * Enabled by the engine demo; defaults to false everywhere else.
      */
     forceAutoApply?: boolean;
+    /**
+     * When false, the review-assistance stage (the per-page eligibility gate plus
+     * the structural review-assistance runner) is skipped entirely and only the
+     * text-correction stage runs. Defaults to true.
+     */
+    reviewAssistanceEnabled?: boolean;
     temperature?: number;
 }
 interface NormalizedPDFCorrectionOptions {
@@ -67,6 +73,7 @@ interface NormalizedPDFCorrectionOptions {
     autoApplyThreshold: number;
     proposalThreshold: number;
     forceAutoApply: boolean;
+    reviewAssistanceEnabled: boolean;
     temperature: number;
 }
 declare const PDF_CORRECTION_DEFAULTS: Omit<NormalizedPDFCorrectionOptions, 'models'>;

package/dist/index.d.ts

@@ -54,6 +54,12 @@ interface PDFCorrectionOptions {
      * Enabled by the engine demo; defaults to false everywhere else.
      */
     forceAutoApply?: boolean;
+    /**
+     * When false, the review-assistance stage (the per-page eligibility gate plus
+     * the structural review-assistance runner) is skipped entirely and only the
+     * text-correction stage runs. Defaults to true.
+     */
+    reviewAssistanceEnabled?: boolean;
     temperature?: number;
 }
 interface NormalizedPDFCorrectionOptions {
@@ -67,6 +73,7 @@ interface NormalizedPDFCorrectionOptions {
     autoApplyThreshold: number;
     proposalThreshold: number;
     forceAutoApply: boolean;
+    reviewAssistanceEnabled: boolean;
     temperature: number;
 }
 declare const PDF_CORRECTION_DEFAULTS: Omit<NormalizedPDFCorrectionOptions, 'models'>;

package/dist/index.js

@@ -590,6 +590,9 @@ var DoclingServer = class _DoclingServer {
     this.venvPath = venvPath;
     this.port = port;
   }
+  logger;
+  venvPath;
+  port;
   async isPortInUse() {
     try {
       const result = await spawnAsync("lsof", ["-ti", `:${this.port}`]);
@@ -736,6 +739,8 @@ var PythonEnvironment = class {
     this.logger = logger;
     this.venvPath = venvPath;
   }
+  logger;
+  venvPath;
   async setup() {
     await this.checkPythonVersion();
     await this.setupPythonEnvironment();
@@ -989,6 +994,7 @@ var PdfTextExtractor = class _PdfTextExtractor {
   constructor(logger) {
     this.logger = logger;
   }
+  logger;
   /**
    * Safely extract page texts: returns the map on success, `undefined` on
    * failure. Failures are logged as a single warning and do not throw, so
@@ -1141,6 +1147,8 @@ var PdfLanguageDetector = class {
     this.pageRenderer = pageRenderer;
     this.textExtractor = textExtractor ?? new PdfTextExtractor(logger);
   }
+  logger;
+  pageRenderer;
   textExtractor;
   async detect(pdfPath, outputDir, options) {
     this.logger.info(
@@ -1357,6 +1365,8 @@ var ImagePdfFallbackError = class extends Error {
     this.originalError = originalError;
     this.fallbackError = fallbackError;
   }
+  originalError;
+  fallbackError;
   name = "ImagePdfFallbackError";
 };
 
@@ -1368,6 +1378,7 @@ var PageRenderer = class {
   constructor(logger) {
     this.logger = logger;
   }
+  logger;
   lastLoggedPercent = 0;
   /**
    * Render all pages of a PDF to individual PNG files.
@@ -1493,6 +1504,7 @@ var InvalidDocumentTypeError = class extends Error {
     );
     this.reason = reason;
   }
+  reason;
   name = "InvalidDocumentTypeError";
   code = "INVALID_DOCUMENT_TYPE";
 };
@@ -2423,6 +2435,10 @@ var ChunkedPDFConverter = class {
     this.config = config;
     this.timeout = timeout;
   }
+  logger;
+  client;
+  config;
+  timeout;
   /**
    * Convert a local PDF in chunks.
    *
@@ -2769,6 +2785,9 @@ var DoclingConversionExecutor = class {
     this.client = client;
     this.timeout = timeout;
   }
+  logger;
+  client;
+  timeout;
   /**
    * Execute a full Docling conversion pipeline.
    *
@@ -2931,6 +2950,7 @@ var ImagePdfConverter = class {
   constructor(logger) {
     this.logger = logger;
   }
+  logger;
   /**
    * Convert a PDF file to an image-based PDF.
    * Downloads the PDF from URL, converts it using ImageMagick, and returns the path.
@@ -3415,6 +3435,7 @@ var PostDoclingPageProcessor = class {
   constructor(logger) {
     this.logger = logger;
   }
+  logger;
   /**
    * Read DoclingDocument from output directory, correct text via VLM,
    * and save the corrected document back.
@@ -5221,6 +5242,8 @@ var ImageRegionSnapper = class {
     this.logger = logger;
     this.options = options;
   }
+  logger;
+  options;
   async snap(pageImagePath, pageSize, bbox) {
     const warnings = [];
     const dims = await this.getImageDimensions(pageImagePath);
@@ -5433,6 +5456,8 @@ var PictureSplitCandidateDetector = class {
     this.options = options;
     this.snapper = snapper ?? new ImageRegionSnapper(logger);
   }
+  logger;
+  options;
   snapper;
   async detect(input) {
     if (!this.isLargeEnoughForInspection(input.pictureBbox)) {
@@ -5734,6 +5759,9 @@ var ReviewAssistanceCheckpointStore = class _ReviewAssistanceCheckpointStore {
     this.checkpoint = checkpoint;
     this.now = now;
   }
+  path;
+  checkpoint;
+  now;
   static open(outputDir, runId, now) {
     const path = join12(outputDir, CHECKPOINT_FILE);
     const checkpoint = this.read(path, runId, now?.() ?? /* @__PURE__ */ new Date());
@@ -6107,6 +6135,7 @@ var ImageCropWriter = class {
     this.logger = logger;
     this.snapper = snapper ?? new ImageRegionSnapper(logger);
   }
+  logger;
   snapper;
   async writeCrop(input) {
     const imagesDir = join13(input.outputDir, "images");
@@ -6229,6 +6258,7 @@ var ReviewAssistancePatcher = class {
     this.snapper = new ImageRegionSnapper(logger);
     this.cropWriter = new ImageCropWriter(logger, this.snapper);
   }
+  logger;
   snapper;
   cropWriter;
   captionResolver = new OrphanCaptionResolver();
@@ -8640,6 +8670,7 @@ var ReviewAssistanceRunner = class {
   constructor(logger) {
     this.logger = logger;
   }
+  logger;
   async analyzeAndSave(outputDir, reportId, modelResolver, options) {
     this.emitProgress(options, {
       substage: "review-assistance:prepare",
@@ -10437,6 +10468,7 @@ var PDF_CORRECTION_DEFAULTS = {
   autoApplyThreshold: 0.85,
   proposalThreshold: 0.5,
   forceAutoApply: false,
+  reviewAssistanceEnabled: true,
   temperature: 0
 };
 function normalizeNumber(value, fallback) {
@@ -10536,6 +10568,7 @@ function normalizePDFCorrectionOptions(value) {
     autoApplyThreshold,
     proposalThreshold,
     forceAutoApply: value.forceAutoApply ?? PDF_CORRECTION_DEFAULTS.forceAutoApply,
+    reviewAssistanceEnabled: value.reviewAssistanceEnabled ?? PDF_CORRECTION_DEFAULTS.reviewAssistanceEnabled,
     temperature: normalizeTemperature(
       value.temperature,
       PDF_CORRECTION_DEFAULTS.temperature
@@ -10548,6 +10581,7 @@ var PostDoclingCorrectionPipeline = class {
   constructor(logger) {
     this.logger = logger;
   }
+  logger;
   wrapCallback(pdfPath, reportId, options, originalCallback, abortSignal) {
     const correction = normalizePDFCorrectionOptions(options.correction);
     return async (outputDir) => {
@@ -10569,44 +10603,49 @@ var PostDoclingCorrectionPipeline = class {
           onTokenUsage: options.onTokenUsage,
           documentLanguages: options.ocr_lang,
           pageTexts,
-          reviewAssistanceGate: {
+          // When review assistance is disabled, omit the gate entirely so the
+          // per-page eligibility checks (pageGate LLM calls + sidecar report)
+          // never run — only text correction executes.
+          reviewAssistanceGate: correction.reviewAssistanceEnabled ? {
             model: correction.models.pageGate,
             fallback: correction.models.pageGateFallback,
             maxRetries: correction.maxRetries.pageGate,
             temperature: correction.temperature,
             outputLanguage: correction.outputLanguage
-          }
-        }
-      );
-      const runner = new ReviewAssistanceRunner(this.logger);
-      await runner.analyzeAndSave(
-        outputDir,
-        reportId,
-        this.buildReviewAssistanceModelResolver(correction),
-        {
-          pdfPath,
-          pageTexts,
-          pageGateModel: correction.models.pageGate,
-          pageGateFallback: correction.models.pageGateFallback,
-          pageGateMaxRetries: correction.maxRetries.pageGate,
-          pageGateTemperature: correction.temperature,
-          pageConcurrency: correction.concurrency.pages,
-          taskConcurrency: correction.concurrency.reviewTasks,
-          modelConcurrency: correction.modelConcurrency,
-          workItemTimeoutMs: correction.workItemTimeoutMs,
-          autoApplyThreshold: correction.autoApplyThreshold,
-          proposalThreshold: correction.proposalThreshold,
-          forceAutoApply: correction.forceAutoApply,
-          maxRetries: correction.maxRetries.reviewAssistance,
-          tableMaxRetries: correction.maxRetries.tableCorrection,
-          temperature: correction.temperature,
-          outputLanguage: correction.outputLanguage,
-          aggregator: options.aggregator,
-          abortSignal,
-          onTokenUsage: options.onTokenUsage,
-          onProgress: options.onReviewAssistanceProgress
+          } : void 0
         }
       );
+      if (correction.reviewAssistanceEnabled) {
+        const runner = new ReviewAssistanceRunner(this.logger);
+        await runner.analyzeAndSave(
+          outputDir,
+          reportId,
+          this.buildReviewAssistanceModelResolver(correction),
+          {
+            pdfPath,
+            pageTexts,
+            pageGateModel: correction.models.pageGate,
+            pageGateFallback: correction.models.pageGateFallback,
+            pageGateMaxRetries: correction.maxRetries.pageGate,
+            pageGateTemperature: correction.temperature,
+            pageConcurrency: correction.concurrency.pages,
+            taskConcurrency: correction.concurrency.reviewTasks,
+            modelConcurrency: correction.modelConcurrency,
+            workItemTimeoutMs: correction.workItemTimeoutMs,
+            autoApplyThreshold: correction.autoApplyThreshold,
+            proposalThreshold: correction.proposalThreshold,
+            forceAutoApply: correction.forceAutoApply,
+            maxRetries: correction.maxRetries.reviewAssistance,
+            tableMaxRetries: correction.maxRetries.tableCorrection,
+            temperature: correction.temperature,
+            outputLanguage: correction.outputLanguage,
+            aggregator: options.aggregator,
+            abortSignal,
+            onTokenUsage: options.onTokenUsage,
+            onProgress: options.onReviewAssistanceProgress
+          }
+        );
+      }
       await originalCallback(outputDir);
     };
   }
@@ -10649,6 +10688,10 @@ var PDFConverter = class {
     this.enableImagePdfFallback = enableImagePdfFallback;
     this.timeout = timeout;
   }
+  logger;
+  client;
+  enableImagePdfFallback;
+  timeout;
   documentTypeValidated = false;
   /**
    * Validate that the PDF is an archaeological investigation report.