@claritylabs/cl-sdk 1.1.2 → 1.2.0

package/dist/index.d.mts

@@ -524,6 +524,20 @@ declare function extractQuoteFromPdf(pdfBase64: string, options?: ExtractOptions
     extracted: any;
 }>;
 
+/**
+ * Extract a page range from a PDF and return as base64.
+ * Used to reduce API token usage by only sending relevant pages.
+ *
+ * @param pdfBase64 - Full PDF as base64 string.
+ * @param startPage - First page to include (1-indexed).
+ * @param endPage - Last page to include (1-indexed, clamped to total pages).
+ * @returns Base64 string of the trimmed PDF, or original if range covers all pages.
+ */
+declare function extractPageRange(pdfBase64: string, startPage: number, endPage: number): Promise<string>;
+/**
+ * Get the page count of a PDF without fully parsing it.
+ */
+declare function getPdfPageCount(pdfBase64: string): Promise<number>;
 interface AcroFormFieldInfo {
     name: string;
     type: "text" | "checkbox" | "dropdown" | "radio";
@@ -548,4 +562,4 @@ interface TextOverlay {
 /** Overlay text on a flat PDF at specified coordinates. */
 declare function overlayTextOnPdf(pdfBytes: Uint8Array, overlays: TextOverlay[]): Promise<Uint8Array>;
 
-export { AGENT_TOOLS, APPLICATION_CLASSIFY_PROMPT, type AcroFormFieldInfo, type AgentContext, type BaseDocument, CLASSIFY_DOCUMENT_PROMPT, CLASSIFY_EMAIL_PROMPT, COI_GENERATION_TOOL, COVERAGE_COMPARISON_TOOL, type ClassifyOptions, type CommunicationIntent, type Coverage, DOCUMENT_LOOKUP_TOOL, EXTRACTION_PROMPT, type ExtractOptions, type ExtractSectionsOptions, type FieldMapping, HAIKU_MODEL, type InsuranceDocument, type LogFn, METADATA_PROMPT, MODEL_TOKEN_LIMITS, type ModelConfig, PLATFORM_CONFIGS, type Platform, type PlatformConfig, type PolicyDocument, type PremiumLine, type PromptBuilder, QUOTE_METADATA_PROMPT, type QuoteDocument, SONNET_MODEL, type Section, type Subjectivity, type Subsection, type TextOverlay, type TokenUsage, type ToolDefinition, type UnderwritingCondition, applyExtracted, applyExtractedQuote, buildAcroFormMappingPrompt, buildAgentSystemPrompt, buildAnswerParsingPrompt, buildAutoFillPrompt, buildBatchEmailGenerationPrompt, buildClassifyMessagePrompt, buildCoiRoutingPrompt, buildConfirmationSummaryPrompt, buildConversationMemoryContext, buildConversationMemoryGuidance, buildCoverageGapPrompt, buildDocumentContext, buildFieldExplanationPrompt, buildFieldExtractionPrompt, buildFlatPdfMappingPrompt, buildFormattingPrompt, buildIdentityPrompt, buildIntentPrompt, buildLookupFillPrompt, buildPolicyContext, buildPolicySectionsPrompt, buildQuestionBatchPrompt, buildQuoteSectionsPrompt, buildQuotesPoliciesPrompt, buildReplyIntentClassificationPrompt, buildSafetyPrompt, buildSectionsPrompt, buildSupplementaryEnrichmentPrompt, buildSystemPrompt, classifyDocumentType, createDefaultModelConfig, createUniformModelConfig, enrichSupplementaryFields, extractFromPdf, extractQuoteFromPdf, extractSectionsOnly, fillAcroForm, getAcroFormFields, getPageChunks, mergeChunkedQuoteSections, mergeChunkedSections, overlayTextOnPdf, sanitizeNulls, stripFences };
+export { AGENT_TOOLS, APPLICATION_CLASSIFY_PROMPT, type AcroFormFieldInfo, type AgentContext, type BaseDocument, CLASSIFY_DOCUMENT_PROMPT, CLASSIFY_EMAIL_PROMPT, COI_GENERATION_TOOL, COVERAGE_COMPARISON_TOOL, type ClassifyOptions, type CommunicationIntent, type Coverage, DOCUMENT_LOOKUP_TOOL, EXTRACTION_PROMPT, type ExtractOptions, type ExtractSectionsOptions, type FieldMapping, HAIKU_MODEL, type InsuranceDocument, type LogFn, METADATA_PROMPT, MODEL_TOKEN_LIMITS, type ModelConfig, PLATFORM_CONFIGS, type Platform, type PlatformConfig, type PolicyDocument, type PremiumLine, type PromptBuilder, QUOTE_METADATA_PROMPT, type QuoteDocument, SONNET_MODEL, type Section, type Subjectivity, type Subsection, type TextOverlay, type TokenUsage, type ToolDefinition, type UnderwritingCondition, applyExtracted, applyExtractedQuote, buildAcroFormMappingPrompt, buildAgentSystemPrompt, buildAnswerParsingPrompt, buildAutoFillPrompt, buildBatchEmailGenerationPrompt, buildClassifyMessagePrompt, buildCoiRoutingPrompt, buildConfirmationSummaryPrompt, buildConversationMemoryContext, buildConversationMemoryGuidance, buildCoverageGapPrompt, buildDocumentContext, buildFieldExplanationPrompt, buildFieldExtractionPrompt, buildFlatPdfMappingPrompt, buildFormattingPrompt, buildIdentityPrompt, buildIntentPrompt, buildLookupFillPrompt, buildPolicyContext, buildPolicySectionsPrompt, buildQuestionBatchPrompt, buildQuoteSectionsPrompt, buildQuotesPoliciesPrompt, buildReplyIntentClassificationPrompt, buildSafetyPrompt, buildSectionsPrompt, buildSupplementaryEnrichmentPrompt, buildSystemPrompt, classifyDocumentType, createDefaultModelConfig, createUniformModelConfig, enrichSupplementaryFields, extractFromPdf, extractPageRange, extractQuoteFromPdf, extractSectionsOnly, fillAcroForm, getAcroFormFields, getPageChunks, getPdfPageCount, mergeChunkedQuoteSections, mergeChunkedSections, overlayTextOnPdf, sanitizeNulls, stripFences };

package/dist/index.d.ts

@@ -524,6 +524,20 @@ declare function extractQuoteFromPdf(pdfBase64: string, options?: ExtractOptions
     extracted: any;
 }>;
 
+/**
+ * Extract a page range from a PDF and return as base64.
+ * Used to reduce API token usage by only sending relevant pages.
+ *
+ * @param pdfBase64 - Full PDF as base64 string.
+ * @param startPage - First page to include (1-indexed).
+ * @param endPage - Last page to include (1-indexed, clamped to total pages).
+ * @returns Base64 string of the trimmed PDF, or original if range covers all pages.
+ */
+declare function extractPageRange(pdfBase64: string, startPage: number, endPage: number): Promise<string>;
+/**
+ * Get the page count of a PDF without fully parsing it.
+ */
+declare function getPdfPageCount(pdfBase64: string): Promise<number>;
 interface AcroFormFieldInfo {
     name: string;
     type: "text" | "checkbox" | "dropdown" | "radio";
@@ -548,4 +562,4 @@ interface TextOverlay {
 /** Overlay text on a flat PDF at specified coordinates. */
 declare function overlayTextOnPdf(pdfBytes: Uint8Array, overlays: TextOverlay[]): Promise<Uint8Array>;
 
-export { AGENT_TOOLS, APPLICATION_CLASSIFY_PROMPT, type AcroFormFieldInfo, type AgentContext, type BaseDocument, CLASSIFY_DOCUMENT_PROMPT, CLASSIFY_EMAIL_PROMPT, COI_GENERATION_TOOL, COVERAGE_COMPARISON_TOOL, type ClassifyOptions, type CommunicationIntent, type Coverage, DOCUMENT_LOOKUP_TOOL, EXTRACTION_PROMPT, type ExtractOptions, type ExtractSectionsOptions, type FieldMapping, HAIKU_MODEL, type InsuranceDocument, type LogFn, METADATA_PROMPT, MODEL_TOKEN_LIMITS, type ModelConfig, PLATFORM_CONFIGS, type Platform, type PlatformConfig, type PolicyDocument, type PremiumLine, type PromptBuilder, QUOTE_METADATA_PROMPT, type QuoteDocument, SONNET_MODEL, type Section, type Subjectivity, type Subsection, type TextOverlay, type TokenUsage, type ToolDefinition, type UnderwritingCondition, applyExtracted, applyExtractedQuote, buildAcroFormMappingPrompt, buildAgentSystemPrompt, buildAnswerParsingPrompt, buildAutoFillPrompt, buildBatchEmailGenerationPrompt, buildClassifyMessagePrompt, buildCoiRoutingPrompt, buildConfirmationSummaryPrompt, buildConversationMemoryContext, buildConversationMemoryGuidance, buildCoverageGapPrompt, buildDocumentContext, buildFieldExplanationPrompt, buildFieldExtractionPrompt, buildFlatPdfMappingPrompt, buildFormattingPrompt, buildIdentityPrompt, buildIntentPrompt, buildLookupFillPrompt, buildPolicyContext, buildPolicySectionsPrompt, buildQuestionBatchPrompt, buildQuoteSectionsPrompt, buildQuotesPoliciesPrompt, buildReplyIntentClassificationPrompt, buildSafetyPrompt, buildSectionsPrompt, buildSupplementaryEnrichmentPrompt, buildSystemPrompt, classifyDocumentType, createDefaultModelConfig, createUniformModelConfig, enrichSupplementaryFields, extractFromPdf, extractQuoteFromPdf, extractSectionsOnly, fillAcroForm, getAcroFormFields, getPageChunks, mergeChunkedQuoteSections, mergeChunkedSections, overlayTextOnPdf, sanitizeNulls, stripFences };
+export { AGENT_TOOLS, APPLICATION_CLASSIFY_PROMPT, type AcroFormFieldInfo, type AgentContext, type BaseDocument, CLASSIFY_DOCUMENT_PROMPT, CLASSIFY_EMAIL_PROMPT, COI_GENERATION_TOOL, COVERAGE_COMPARISON_TOOL, type ClassifyOptions, type CommunicationIntent, type Coverage, DOCUMENT_LOOKUP_TOOL, EXTRACTION_PROMPT, type ExtractOptions, type ExtractSectionsOptions, type FieldMapping, HAIKU_MODEL, type InsuranceDocument, type LogFn, METADATA_PROMPT, MODEL_TOKEN_LIMITS, type ModelConfig, PLATFORM_CONFIGS, type Platform, type PlatformConfig, type PolicyDocument, type PremiumLine, type PromptBuilder, QUOTE_METADATA_PROMPT, type QuoteDocument, SONNET_MODEL, type Section, type Subjectivity, type Subsection, type TextOverlay, type TokenUsage, type ToolDefinition, type UnderwritingCondition, applyExtracted, applyExtractedQuote, buildAcroFormMappingPrompt, buildAgentSystemPrompt, buildAnswerParsingPrompt, buildAutoFillPrompt, buildBatchEmailGenerationPrompt, buildClassifyMessagePrompt, buildCoiRoutingPrompt, buildConfirmationSummaryPrompt, buildConversationMemoryContext, buildConversationMemoryGuidance, buildCoverageGapPrompt, buildDocumentContext, buildFieldExplanationPrompt, buildFieldExtractionPrompt, buildFlatPdfMappingPrompt, buildFormattingPrompt, buildIdentityPrompt, buildIntentPrompt, buildLookupFillPrompt, buildPolicyContext, buildPolicySectionsPrompt, buildQuestionBatchPrompt, buildQuoteSectionsPrompt, buildQuotesPoliciesPrompt, buildReplyIntentClassificationPrompt, buildSafetyPrompt, buildSectionsPrompt, buildSupplementaryEnrichmentPrompt, buildSystemPrompt, classifyDocumentType, createDefaultModelConfig, createUniformModelConfig, enrichSupplementaryFields, extractFromPdf, extractPageRange, extractQuoteFromPdf, extractSectionsOnly, fillAcroForm, getAcroFormFields, getPageChunks, getPdfPageCount, mergeChunkedQuoteSections, mergeChunkedSections, overlayTextOnPdf, sanitizeNulls, stripFences };

package/dist/index.js

@@ -28532,11 +28532,13 @@ __export(index_exports, {
   createUniformModelConfig: () => createUniformModelConfig,
   enrichSupplementaryFields: () => enrichSupplementaryFields,
   extractFromPdf: () => extractFromPdf,
+  extractPageRange: () => extractPageRange,
   extractQuoteFromPdf: () => extractQuoteFromPdf,
   extractSectionsOnly: () => extractSectionsOnly,
   fillAcroForm: () => fillAcroForm,
   getAcroFormFields: () => getAcroFormFields,
   getPageChunks: () => getPageChunks,
+  getPdfPageCount: () => getPdfPageCount,
   mergeChunkedQuoteSections: () => mergeChunkedQuoteSections,
   mergeChunkedSections: () => mergeChunkedSections,
   overlayTextOnPdf: () => overlayTextOnPdf,
@@ -29862,6 +29864,124 @@ var AGENT_TOOLS = [
 
 // src/extraction/pipeline.ts
 var import_ai = require("ai");
+
+// src/extraction/pdf.ts
+var import_pdf_lib = require("pdf-lib");
+async function extractPageRange(pdfBase64, startPage, endPage) {
+  const srcBytes = typeof Buffer !== "undefined" ? Buffer.from(pdfBase64, "base64") : Uint8Array.from(atob(pdfBase64), (c) => c.charCodeAt(0));
+  const srcDoc = await import_pdf_lib.PDFDocument.load(srcBytes, { ignoreEncryption: true });
+  const totalPages = srcDoc.getPageCount();
+  const start = Math.max(startPage - 1, 0);
+  const end = Math.min(endPage, totalPages) - 1;
+  if (start === 0 && end >= totalPages - 1) {
+    return pdfBase64;
+  }
+  const newDoc = await import_pdf_lib.PDFDocument.create();
+  const indices = Array.from({ length: end - start + 1 }, (_, i) => start + i);
+  const pages = await newDoc.copyPages(srcDoc, indices);
+  pages.forEach((page) => newDoc.addPage(page));
+  const bytes = await newDoc.save();
+  if (typeof Buffer !== "undefined") {
+    return Buffer.from(bytes).toString("base64");
+  }
+  let binary = "";
+  const uint8 = new Uint8Array(bytes);
+  for (let i = 0; i < uint8.length; i++) {
+    binary += String.fromCharCode(uint8[i]);
+  }
+  return btoa(binary);
+}
+async function getPdfPageCount(pdfBase64) {
+  const srcBytes = typeof Buffer !== "undefined" ? Buffer.from(pdfBase64, "base64") : Uint8Array.from(atob(pdfBase64), (c) => c.charCodeAt(0));
+  const doc = await import_pdf_lib.PDFDocument.load(srcBytes, { ignoreEncryption: true });
+  return doc.getPageCount();
+}
+function getAcroFormFields(pdfDoc) {
+  const form = pdfDoc.getForm();
+  const fields = form.getFields();
+  if (fields.length === 0) return [];
+  return fields.map((field) => {
+    const name = field.getName();
+    if (field instanceof import_pdf_lib.PDFTextField) {
+      return { name, type: "text" };
+    }
+    if (field instanceof import_pdf_lib.PDFCheckBox) {
+      return { name, type: "checkbox" };
+    }
+    if (field instanceof import_pdf_lib.PDFDropdown) {
+      return { name, type: "dropdown", options: field.getOptions() };
+    }
+    if (field instanceof import_pdf_lib.PDFRadioGroup) {
+      return { name, type: "radio", options: field.getOptions() };
+    }
+    return { name, type: "text" };
+  });
+}
+async function fillAcroForm(pdfBytes, mappings) {
+  const pdfDoc = await import_pdf_lib.PDFDocument.load(pdfBytes, { ignoreEncryption: true });
+  const form = pdfDoc.getForm();
+  for (const { acroFormName, value } of mappings) {
+    try {
+      const field = form.getField(acroFormName);
+      if (field instanceof import_pdf_lib.PDFTextField) {
+        field.setText(value);
+      } else if (field instanceof import_pdf_lib.PDFCheckBox) {
+        const lower = value.toLowerCase();
+        if (["yes", "true", "x", "checked", "on"].includes(lower)) {
+          field.check();
+        } else {
+          field.uncheck();
+        }
+      } else if (field instanceof import_pdf_lib.PDFDropdown) {
+        try {
+          field.select(value);
+        } catch {
+        }
+      } else if (field instanceof import_pdf_lib.PDFRadioGroup) {
+        try {
+          field.select(value);
+        } catch {
+        }
+      }
+    } catch {
+    }
+  }
+  form.flatten();
+  return await pdfDoc.save();
+}
+async function overlayTextOnPdf(pdfBytes, overlays) {
+  const pdfDoc = await import_pdf_lib.PDFDocument.load(pdfBytes, { ignoreEncryption: true });
+  const font = await pdfDoc.embedFont(import_pdf_lib.StandardFonts.Helvetica);
+  const pageCount = pdfDoc.getPageCount();
+  for (const overlay of overlays) {
+    if (overlay.page < 0 || overlay.page >= pageCount) continue;
+    const page = pdfDoc.getPage(overlay.page);
+    const { width, height } = page.getSize();
+    const fontSize = overlay.fontSize ?? 10;
+    const x = overlay.x / 100 * width;
+    const y = height - overlay.y / 100 * height - fontSize;
+    if (overlay.isCheckmark) {
+      page.drawText("X", {
+        x,
+        y,
+        size: fontSize,
+        font,
+        color: (0, import_pdf_lib.rgb)(0, 0, 0)
+      });
+    } else {
+      page.drawText(overlay.text, {
+        x,
+        y,
+        size: fontSize,
+        font,
+        color: (0, import_pdf_lib.rgb)(0, 0, 0)
+      });
+    }
+  }
+  return await pdfDoc.save();
+}
+
+// src/extraction/pipeline.ts
 var SONNET_MODEL = "claude-sonnet-4-6";
 var HAIKU_MODEL = "claude-haiku-4-5-20251001";
 var DEFAULT_METADATA_PROVIDER_OPTIONS = {
@@ -29998,8 +30118,10 @@ function getPageChunks(totalPages, chunkSize = 30) {
   }
   return chunks;
 }
-async function callModel(model, pdfBase64, prompt, maxTokens, providerOptions, log, onTokenUsage) {
-  await log?.(`Calling model (max ${maxTokens} tokens)...`);
+async function callModel(model, pdfBase64, prompt, maxTokens, providerOptions, log, onTokenUsage, pageRange) {
+  const pdfToSend = pageRange ? await extractPageRange(pdfBase64, pageRange[0], pageRange[1]) : pdfBase64;
+  const rangeLabel = pageRange ? ` [pages ${pageRange[0]}\u2013${pageRange[1]}]` : "";
+  await log?.(`Calling model (max ${maxTokens} tokens)${rangeLabel}...`);
   const start = Date.now();
   const { text, usage } = await withRetry(
     () => (0, import_ai.generateText)({
@@ -30008,7 +30130,7 @@ async function callModel(model, pdfBase64, prompt, maxTokens, providerOptions, l
       messages: [{
         role: "user",
         content: [
-          { type: "file", data: pdfBase64, mediaType: "application/pdf" },
+          { type: "file", data: pdfToSend, mediaType: "application/pdf" },
           { type: "text", text: prompt }
         ]
       }],
@@ -30114,7 +30236,9 @@ async function classifyDocumentType(pdfBase64, options) {
     MODEL_TOKEN_LIMITS.classification,
     void 0,
     log,
-    onTokenUsage
+    onTokenUsage,
+    [1, 3]
+    // Only need first 3 pages for classification
   );
   try {
     const parsed = JSON.parse(stripFences(raw));
@@ -30203,7 +30327,9 @@ async function extractChunkWithRetry(models, pdfBase64, start, end, sizeIndex, p
     MODEL_TOKEN_LIMITS.sections,
     void 0,
     log,
-    onTokenUsage
+    onTokenUsage,
+    [start, end]
+    // Only send this chunk's pages
   );
   try {
     return [JSON.parse(stripFences(chunkRaw))];
@@ -30245,7 +30371,9 @@ async function extractChunkWithRetry(models, pdfBase64, start, end, sizeIndex, p
       MODEL_TOKEN_LIMITS.sectionsFallback,
       fallbackProviderOptions,
       log,
-      onTokenUsage
+      onTokenUsage,
+      [start, end]
+      // Only send this chunk's pages
     );
     try {
       return [JSON.parse(stripFences(fallbackRaw))];
@@ -30288,7 +30416,9 @@ async function extractFromPdf(pdfBase64, options) {
     onTokenUsage
   } = options ?? {};
   const resolved = resolveModels(models);
+  const actualPageCount = await getPdfPageCount(pdfBase64);
   await log?.("Pass 1: Extracting metadata...");
+  const metadataPageRange = [1, Math.min(10, actualPageCount)];
   const metadataRaw = await callModel(
     resolved.metadata,
     pdfBase64,
@@ -30296,7 +30426,8 @@ async function extractFromPdf(pdfBase64, options) {
     MODEL_TOKEN_LIMITS.metadata,
     metadataProviderOptions,
     log,
-    onTokenUsage
+    onTokenUsage,
+    metadataPageRange
   );
   let metadataResult;
   try {
@@ -30307,7 +30438,7 @@ async function extractFromPdf(pdfBase64, options) {
     throw new Error(`Metadata JSON parse failed: ${e.message}`);
   }
   await onMetadata?.(metadataRaw);
-  const pageCount = metadataResult.totalPages || 1;
+  const pageCount = actualPageCount;
   await log?.(`Document: ${pageCount} page(s)`);
   const sectionChunks = await extractSectionChunks(
     resolved,
@@ -30375,7 +30506,9 @@ async function extractQuoteFromPdf(pdfBase64, options) {
     onTokenUsage
   } = options ?? {};
   const resolved = resolveModels(models);
+  const actualPageCount = await getPdfPageCount(pdfBase64);
   await log?.("Pass 1: Extracting quote metadata...");
+  const metadataPageRange = [1, Math.min(10, actualPageCount)];
   const metadataRaw = await callModel(
     resolved.metadata,
     pdfBase64,
@@ -30383,7 +30516,8 @@ async function extractQuoteFromPdf(pdfBase64, options) {
     MODEL_TOKEN_LIMITS.metadata,
     metadataProviderOptions,
     log,
-    onTokenUsage
+    onTokenUsage,
+    metadataPageRange
   );
   let metadataResult;
   try {
@@ -30394,7 +30528,7 @@ async function extractQuoteFromPdf(pdfBase64, options) {
     throw new Error(`Quote metadata JSON parse failed: ${e.message}`);
   }
   await onMetadata?.(metadataRaw);
-  const pageCount = metadataResult.totalPages || 1;
+  const pageCount = actualPageCount;
   await log?.(`Quote document: ${pageCount} page(s)`);
   const sectionChunks = await extractSectionChunks(
     resolved,
@@ -30411,93 +30545,6 @@ async function extractQuoteFromPdf(pdfBase64, options) {
   const mergedRaw = JSON.stringify(merged);
   return { rawText: mergedRaw, extracted: merged };
 }
-
-// src/extraction/pdf.ts
-var import_pdf_lib = require("pdf-lib");
-function getAcroFormFields(pdfDoc) {
-  const form = pdfDoc.getForm();
-  const fields = form.getFields();
-  if (fields.length === 0) return [];
-  return fields.map((field) => {
-    const name = field.getName();
-    if (field instanceof import_pdf_lib.PDFTextField) {
-      return { name, type: "text" };
-    }
-    if (field instanceof import_pdf_lib.PDFCheckBox) {
-      return { name, type: "checkbox" };
-    }
-    if (field instanceof import_pdf_lib.PDFDropdown) {
-      return { name, type: "dropdown", options: field.getOptions() };
-    }
-    if (field instanceof import_pdf_lib.PDFRadioGroup) {
-      return { name, type: "radio", options: field.getOptions() };
-    }
-    return { name, type: "text" };
-  });
-}
-async function fillAcroForm(pdfBytes, mappings) {
-  const pdfDoc = await import_pdf_lib.PDFDocument.load(pdfBytes, { ignoreEncryption: true });
-  const form = pdfDoc.getForm();
-  for (const { acroFormName, value } of mappings) {
-    try {
-      const field = form.getField(acroFormName);
-      if (field instanceof import_pdf_lib.PDFTextField) {
-        field.setText(value);
-      } else if (field instanceof import_pdf_lib.PDFCheckBox) {
-        const lower = value.toLowerCase();
-        if (["yes", "true", "x", "checked", "on"].includes(lower)) {
-          field.check();
-        } else {
-          field.uncheck();
-        }
-      } else if (field instanceof import_pdf_lib.PDFDropdown) {
-        try {
-          field.select(value);
-        } catch {
-        }
-      } else if (field instanceof import_pdf_lib.PDFRadioGroup) {
-        try {
-          field.select(value);
-        } catch {
-        }
-      }
-    } catch {
-    }
-  }
-  form.flatten();
-  return await pdfDoc.save();
-}
-async function overlayTextOnPdf(pdfBytes, overlays) {
-  const pdfDoc = await import_pdf_lib.PDFDocument.load(pdfBytes, { ignoreEncryption: true });
-  const font = await pdfDoc.embedFont(import_pdf_lib.StandardFonts.Helvetica);
-  const pageCount = pdfDoc.getPageCount();
-  for (const overlay of overlays) {
-    if (overlay.page < 0 || overlay.page >= pageCount) continue;
-    const page = pdfDoc.getPage(overlay.page);
-    const { width, height } = page.getSize();
-    const fontSize = overlay.fontSize ?? 10;
-    const x = overlay.x / 100 * width;
-    const y = height - overlay.y / 100 * height - fontSize;
-    if (overlay.isCheckmark) {
-      page.drawText("X", {
-        x,
-        y,
-        size: fontSize,
-        font,
-        color: (0, import_pdf_lib.rgb)(0, 0, 0)
-      });
-    } else {
-      page.drawText(overlay.text, {
-        x,
-        y,
-        size: fontSize,
-        font,
-        color: (0, import_pdf_lib.rgb)(0, 0, 0)
-      });
-    }
-  }
-  return await pdfDoc.save();
-}
 // Annotate the CommonJS export names for ESM import in node:
 0 && (module.exports = {
   AGENT_TOOLS,
@@ -30550,11 +30597,13 @@ async function overlayTextOnPdf(pdfBytes, overlays) {
   createUniformModelConfig,
   enrichSupplementaryFields,
   extractFromPdf,
+  extractPageRange,
   extractQuoteFromPdf,
   extractSectionsOnly,
   fillAcroForm,
   getAcroFormFields,
   getPageChunks,
+  getPdfPageCount,
   mergeChunkedQuoteSections,
   mergeChunkedSections,
   overlayTextOnPdf,