@heripo/pdf-parser 0.1.38 → 0.1.40

package/dist/index.d.ts

@@ -60,6 +60,13 @@ interface PDFCorrectionOptions {
      * text-correction stage runs. Defaults to true.
      */
     reviewAssistanceEnabled?: boolean;
+    /**
+     * When false, the table-correction work item is not generated during review
+     * assistance scheduling. The rest of the review-assistance pipeline (text
+     * OCR, text integrity, footnotes, pictures, layout) continues to run. Only
+     * relevant when `reviewAssistanceEnabled` is true. Defaults to true.
+     */
+    tableCorrectionEnabled?: boolean;
     temperature?: number;
 }
 interface NormalizedPDFCorrectionOptions {
@@ -74,6 +81,7 @@ interface NormalizedPDFCorrectionOptions {
     proposalThreshold: number;
     forceAutoApply: boolean;
     reviewAssistanceEnabled: boolean;
+    tableCorrectionEnabled: boolean;
     temperature: number;
 }
 declare const PDF_CORRECTION_DEFAULTS: Omit<NormalizedPDFCorrectionOptions, 'models'>;

package/dist/index.js

@@ -64,7 +64,7 @@ var IMAGE_PDF_CONVERTER = {
   QUALITY: 100
 };
 
-// ../shared/dist/index.mjs
+// ../shared/dist/index.js
 import { spawn } from "child_process";
 import {
   NoObjectGeneratedError,
@@ -309,7 +309,7 @@ var LLMCaller = class {
     return this.executeWithFallback(
       config,
       (model) => this.generateStructuredOutput(model, config.schema, {
-        system: config.systemPrompt,
+        instructions: config.systemPrompt,
         prompt: config.userPrompt,
         temperature: config.temperature,
         maxRetries: config.maxRetries,
@@ -1328,8 +1328,8 @@ var PdfLanguageDetector = class {
               text: KOREAN_DOCUMENT_DETECTION_PROMPT
             },
             {
-              type: "image",
-              image: imageData,
+              type: "file",
+              data: imageData,
               mediaType: "image/png"
             }
           ]
@@ -3309,8 +3309,8 @@ var ReviewAssistancePageGate = class {
               text: this.buildPrompt(context, options.outputLanguage)
             },
             {
-              type: "image",
-              image,
+              type: "file",
+              data: image,
               mediaType: "image/png"
             }
           ]
@@ -3582,8 +3582,8 @@ var PostDoclingPageProcessor = class {
                 text: fullPrompt
               },
               {
-                type: "image",
-                image,
+                type: "file",
+                data: image,
                 mediaType: "image/png"
               }
             ]
@@ -7943,17 +7943,18 @@ var TEXT_ROLE_REASONS = /* @__PURE__ */ new Set([
 ]);
 var SEVERE_TABLE_REASONS = /* @__PURE__ */ new Set(["multi_page_table_candidate"]);
 var ReviewAssistanceWorkScheduler = class {
-  build(context) {
+  build(context, options) {
     if (!context.reviewAssistanceEligibility.eligible) return [];
-    return [
+    const items = [
       ...this.buildTextOcrItems(context),
       ...this.buildTextIntegrityItems(context),
       ...this.buildTextRoleItems(context),
-      ...this.buildTableItems(context),
+      ...options?.skipTableItems ? [] : this.buildTableItems(context),
       ...this.buildPictureCaptionItems(context),
       ...this.buildPictureSplitItems(context),
       ...this.buildLayoutItems(context)
     ];
+    return items;
   }
   buildTextOcrItems(context) {
     const targetRefs = context.textBlocks.filter(
@@ -9016,7 +9017,9 @@ var ReviewAssistanceRunner = class {
       );
       const image = new Uint8Array(await readFile(context.pageImagePath));
       const scheduler = new ReviewAssistanceWorkScheduler();
-      const workItems = scheduler.build(context);
+      const workItems = scheduler.build(context, {
+        skipTableItems: !options.tableCorrectionEnabled
+      });
       if (workItems.length === 0) {
         const result = {
           pageNo: context.pageNo,
@@ -9204,8 +9207,8 @@ var ReviewAssistanceRunner = class {
                 content: [
                   { type: "text", text: prompt },
                   {
-                    type: "image",
-                    image,
+                    type: "file",
+                    data: image,
                     mediaType: "image/png"
                   }
                 ]
@@ -10030,8 +10033,8 @@ var ReviewAssistanceRunner = class {
             content: [
               { type: "text", text: prompt },
               {
-                type: "image",
-                image,
+                type: "file",
+                data: image,
                 mediaType: "image/png"
               }
             ]
@@ -10469,6 +10472,7 @@ var PDF_CORRECTION_DEFAULTS = {
   proposalThreshold: 0.5,
   forceAutoApply: false,
   reviewAssistanceEnabled: true,
+  tableCorrectionEnabled: true,
   temperature: 0
 };
 function normalizeNumber(value, fallback) {
@@ -10569,6 +10573,7 @@ function normalizePDFCorrectionOptions(value) {
     proposalThreshold,
     forceAutoApply: value.forceAutoApply ?? PDF_CORRECTION_DEFAULTS.forceAutoApply,
     reviewAssistanceEnabled: value.reviewAssistanceEnabled ?? PDF_CORRECTION_DEFAULTS.reviewAssistanceEnabled,
+    tableCorrectionEnabled: value.tableCorrectionEnabled ?? PDF_CORRECTION_DEFAULTS.tableCorrectionEnabled,
     temperature: normalizeTemperature(
       value.temperature,
       PDF_CORRECTION_DEFAULTS.temperature
@@ -10637,6 +10642,7 @@ var PostDoclingCorrectionPipeline = class {
             forceAutoApply: correction.forceAutoApply,
             maxRetries: correction.maxRetries.reviewAssistance,
             tableMaxRetries: correction.maxRetries.tableCorrection,
+            tableCorrectionEnabled: correction.tableCorrectionEnabled,
             temperature: correction.temperature,
             outputLanguage: correction.outputLanguage,
             aggregator: options.aggregator,